DE3733674A1 - Speech analyser - Google Patents

Abstract

A dictionary analyser has a dictionary device with dictionary data stored in it, including morpheme data for words, composed words and phrases and an analysing device for carrying out a morphological analysis for an entered sentence, making reference to the dictionary device, the dictionary device containing data for the degree of coupling, which indicates the degree of coupling between the respective words which constitute the composed words or phrases, the analysing device referring to the dictionary device for the respective words which are contained in the entered sentence, and if a number of dictionary data are called up for a word in combination with other words, the combination of words having a higher degree of coupling is selected by referring to the data for the degree of coupling.

Description

The invention relates to a speech analyzer and relates to in particular a speech analyzer for analyzing natural ones Languages, the analyzer in conjunction with automatic translation devices can be used should.

With conventional speech analyzers, there are different Difficulties listed below. At the Analyze morphemes in a sentence, such as Words etc., it is important and has a significant impact on the result of the analysis to assess whether a particular Word is used alone, or whether it is a compound Word or phrase or phrase used being coupled with another word or words is.

The conventional analyzers use a system to measure one Perform analysis for both cases, assuming is that successive words are a phrase as well are independent words, and finally a corresponding one Translation is chosen based on this result, or a system is used to preferably choose the successive ones Words as a phrase or phrase judge. The system listed first requires this when processing a long time, while at the second listed system the possibility is great that there is an incorrect translation.

In the morphological analysis carried out during a translation is carried out becomes part of the language or a similar one other information for the morphemes like a word obtained by calling up a dictionary. In the event of  from common words, such as nouns and verbs them since most of them are stored in the dictionary are easily found so as to provide information receive.

However, since expressions such as a length (m), a Speed (m / s), acceleration (m / s²) and the like show other units, extremely versatile meanings it is not efficient to have all of these expressions to save in the dictionary, because of this the storage capacity for dictionary information in uneconomical Way is increased.

In the case of a morphological analysis of a sentence can be the numerical expression in a particular Language is not always exactly the same as in others Languages match. For example, that's basic Concept of counting numbers, d. H. their job notation between the Japanese and European languages, for example the English or German language, different. For example, if an English or German expression for numerical values namely "one-hundred and two thousand two hundred and four" simply in its components is disassembled and only by appropriate expressions in the Japanese language is replaced, then they are only considered "100 and 2,200 and 4" analyzed. This should be correct ultimately as "102.204", i.e. H. in the Japanese language are translated, in which "one" 10 thousand, "sen" thousand and "hyaku" means a hundred. In the morphological analysis, which was carried out during a translation an entered sentence with the help of delimitations, such as space, Comma and colon, in dictionary reference units divided, and in the dictionary they are divided by the dictionary reference units found again, so one Get part of the language and other information. In In this case, for example, a proper name may be dependent from the context in two or more different  Meanings are used. For example "Osaka City" used a group as the subject to express what is stated as "Osaka City ...", and it's also used in meaning to the place to express it as an object, like "... Osaka City". There only one meaning saved for each proper name is, depending on the meanings, not with such various uses be, thereby increasing the accuracy of the morphological analysis is reduced.

For example, in the case of an English / German Sentence. . . in Central Park, John Willson had a (a). . "the sentence was not analyzed morphologically accordingly if it is not recognized that he is in the Proper name "Central Park" and the other proper names "John Willson" is separated in this context. The same Wise must in the case of an English / German sentence. . . in Boston was (what) Mr. Baker. . . be recognized that he thereby in a proper name "Boston" and a proper name "Mr. Baker" is to be divided. In the conventional system however, a number of these proper names are erroneously called Whole recognized as a single proper name.

For example, when analyzing the morphemes, one English sentence a sequence of words, which with a capital letters begin, generally as a proper name recognized. However, if words with capital letters start, persist, it is not always appropriate to consider it To recognize the whole as a proper name. There is one such a case where they actually have a number of proper names are which appear randomly one after the other. In which Analyze morphemes of what translates is performed, a language part and other information obtained by retrieving a dictionary. There in in this case most common nouns, verbs  etc. can be stored in the dictionary, they can can be easily found to the appropriate information to obtain.

However, since versatile types of words for proper names are present, it is not possible to put all of them in to save a dictionary. So these proper names, who are not registered in the dictionary too cannot be recognized as a proper name.

In the event that a string of characters in a very specific Pattern is summarized, is present, is the Great possibility that there is an incorrect grammatical Dissection comes when processing for normal word units applied in common sentences be applied in such a string what may result in a meaningless translation.

For example, in a case where an English String "Sunday, Jan. 26, '80" reads through the a specific date is expressed, very difficult to get this translate into Japanese because this is expressed in Japanese should be called "'80 -nen, 1-gatsu, 26-nichi, nichi yobi ", where" nen "year," gatsu "month," nichi "day and "nichiyobi" mean Sunday. As further for derived words no assessment for the language part and for a semantic Feature has been performed, depending on no exact translation can be obtained for the respective case.

The grammar used to parse a language a context-free grammar, for example (cfg) has the disadvantage that many complex solutions are spent which are then ultimately not used can be, even if a descending (top-down) or an ascending (bottem-up) method is used. Lots these elaborate solutions are quite obvious as  Errors recognized when they are read in practice. There however, a structural transformation or a translation too is carried out for such elaborate solutions, and then the adequacy of the outcome in each Processing steps are assessed, this often runs for an uneconomical processing time. For example the English word "let" has the meaning of a Command and an invitation and consequently a grammatical (morphological or syntactic) analysis for the respective possibilities are carried out, whereby the Efficiency is reduced. Furthermore, it is difficult choose one of the meanings.

You can also speak in English or German a number of words are combined with hyphens, such as "take care of him" ("take-care-of-him-attitude") in order to free one prepare adjective group. However, it is difficult this, as usual, through an ordinary grammatical Treat analysis.

The situation is also for the additional question (tag question) the same. Although the form of the additional question in the English Language is extremely limited, it requires an extreme complicated processing with the usual analysis method. Furthermore, it is not easy to determine which verb to which Additional question is related.

By means of the invention, the above difficulties should be solved be eliminated. A first object of the invention is therefore, to create a speech analyzer that is of scale the degree of coupling between successive words can judge and who can judge whether it is based on this result are a phrase or not. A second object of the invention is a speech analyzer to create a morphological analysis on a entered sentence, which is a compound  String, such as sizing units, without all such strings in stored in a dictionary. A third object of the invention is to create a speech analyzer which an appropriate morphological analysis regarding a Expression that contains numeric values. A fourth object of the invention is a speech analyzer to create which corresponds to a proper name can translate the context into a corresponding meaning. A fifth object of the invention is a speech analyzer to create what a morphological analyzing can correctly perform regarding an expression which contains a number of consecutive nouns. A sixth object of the invention is a speech analyzer to create which is an unregistered proper name can perform and perform an appropriate analysis for you Proper names can perform a relationship between Phrases taken into account before and after are arranged. A seventh object of the invention is to create a speech analyzer regarding morphemes a string can correctly analyze which express a special meaning by using a special rule are combined. An eighth object of the invention is to create a speech analyzer that is a grammatical feature, meaning etc. of a recognized one Word can estimate that a derivative accordingly a predetermined rule. A ninth aim of the invention is it, after all, a speech analyzer too create which is the structural feature of an entered Recognizes sentence and a grammatical analysis accordingly performs this feature.

The above nine objectives are in accordance with the invention (in order of the specified nine destinations) by a speech analyzer according to the features in the claims 1, 3, 8, 10, 16, 17, 21, 23 and 24 solved. The on  the cited claims are directly or indirectly related Subclaims are advantageous according to the invention Further training and additions.

A speech analyzer according to the Invention based on nine different embodiments with reference to the accompanying drawings in detail explained. Show it:

According, wherein FIG. 1 to 10 show a first embodiment of a speech analyzer of the invention, which is applied to an automatic transmission device English- Japanese in detail

Fig. 1 is a functional block diagram of an example of the detailed configuration of a morphological analysis section;

Fig. 2 is a functional diagram of the entire structure;

Figure 3 is an explanatory view, the basis of which an example of the construction of a dictionary file is shown, which is provided with a highest Vorzugsflag.

Fig. 4 is a flowchart showing an example of a morphological analysis;

5 is a flowchart of an example of the input processing in the morphological analysis.

Fig. 6 is an explanatory overview of an example of shaping an input character string;

Fig. 7 is an explanatory diagram showing an example of dictionary retrieval;

Figs. 8A to 8D are flow charts which illustrate an example, is eliminated as a contradiction at the highest Vorzugsflag in the morphological analysis;

Fig. 9 is an explanatory overview of an example of the content of a buffer for retrieved dictionary information after referencing the dictionary;

Fig. 10 is an explanatory diagram showing an example of the contents of the retrieved dictionary information buffer as a result of performing a conflict of highest priority flag;

Fig. 11 to 16 according to a second embodiment, in which showing of the invention

FIG. 11 is a block diagram of the embodiment;

Fig. 12 is a table showing an example of data stored in a dictionary file;

FIG. 13 is a table in which is shown an example of data stored in a dictionary serving as basic unit file;

Fig. 14 is a table showing an example of data stored in a dictionary information preservation table;

FIG. 15 is a flowchart of the operation of this device;

FIG. 16 is a flowchart of the recognition unit;

To 29 a third embodiment of a speech analyzer which is Fig. 17 according to the invention is applied to an automatic transmission device English-Japanese, and show

17 is a functional block diagram of an example of the detailed structure of a morphological analysis;

18 is a functional block diagram of the entire structure;

FIG. 19A and 19B are flowcharts of an example of a morphological analysis;

FIG. 20 is a flowchart showing an example of a collective arrangement is shown for a currency symbol, and for the unit in the morphological analysis;

FIG. 21A and 21B are flowcharts of an example of dashes provided with digits in the morphological analysis;

FIG. 22A and 22B are flowcharts of an example of processing of consecutive numbers in the morphological analysis;

FIG. 23A and 23B are flowcharts of an example of a collective arrangement with a previous numerical value in the morphological analysis;

Fig. 24 is an explanatory diagram showing an example of the structure of a dictionary file provided with a numeric flag;

Fig. 25 is an explanatory view of an example of an input character string;

FIG. 26A to 26D explanatory overviews, in which the content of the dictionary information preservation table is shown, which has been, found in the dictionary for the input character string shown in Figure 25 as a function of the processing steps.

Figure 27 is an explanatory view of another example of the input character string.

Fig. 28 is an explanatory diagram showing the contents of a currency symbol table, a position table and a decimal point table in the dictionary;

FIG. 29A to 29D are explanatory views illustrating an example of the dictionary information preservation table is shown, which has been, found in the dictionary for the input character string shown in Figure 27 in accordance with the processing steps.

Figure 30 to 36 a fourth embodiment. A speech analyzer according to the invention, in which show

FIG. 30 is a block diagram of this embodiment;

Fig. 31 is a table showing an example of data stored in another dictionary;

FIG. 32 is a flowchart of the operation of the entire device;

Fig. 33 is a flowchart of processing for the proper name registered in a dictionary;

Fig. 34 is a flowchart of processing for a proper name not registered in the dictionary;

FIG. 35 is a flowchart of processing for defective feature information;

FIG. 36 is a table of an example in which data-information-preserving dictionary table stored in the changed after the processing for the input sentence;

37 to 46 a fifth embodiment of a speech analyzer which is Fig. According to the invention applied to an automatic transmission device for English-Japanese, and show

FIG. 37 is a functional block diagram of an example of a detailed construction for the morphological analysis section;

Fig. 38 is a functional block diagram of the whole structure;

FIG. 39 is an explanatory view of an embodiment for building a dictionary file;

FIG. 40 is a flowchart showing an example of a morphological analysis for a proper name;

FIG. 41 is a flowchart showing an example of the collective arrangement of proper names, which are registered in a dictionary for the morphological analysis;

Fig. 42 to 44 are flow charts of an example of processing in response to the position information in the morphological analysis of a proper name;

45 is a flowchart showing an example of the collective arrangement for proper names, that are not registered in a dictionary, in the morphological analysis.

FIG. 46A to 46F are explanatory tables, which information-preserving dictionary table representing the contents of the referenced in the dictionary, for example, an input character string with respect to the processing steps is correspondingly taken;

Fig. 47 to 52, a sixth embodiment of a speech analyzer according to the invention, in which show

Fig. 47 is a block diagram of this embodiment;

FIG. 48 is a table example of data stored in a reference dictionary;

Fig. 49 is a flow chart of the operation of the entire device;

Fig. 50 is a flowchart of processing for the proper names stored in the dictionary;

Fig. 51 is a flowchart of processing for proper names that are not registered in the dictionary;

, In which are by way of example data stored in a dictionary-information preservation table of the input sentence changed Fig. 51 is a table

53 to 57, a seventh embodiment of a speech analyzer which is Fig. According to the invention used in conjunction with an automatic transmission device English-Japanese, wherein said

FIG. 53 is a functional block diagram of one embodiment for the detailed structure of the morphological analysis section;

54 is a functional block diagram of the entire structure;

FIG. 55A and 55B are flowcharts of an example of the morphological analysis,

Fig. 56 is an explanatory diagram showing an example of the content of the information table in an information processing section;

Fig. 57 is an explanatory overview of an example of the content of an adjustment table (7128);

Fig. 58 to 63, an eighth embodiment of a speech analyzer according to the invention, in which show

FIG. 58 is a block diagram for explaining the entire structure;

FIG. 59 is a block diagram for explaining an example of processing of words derived by means of a prefix;

FIG. 60 is a block diagram for explaining an example of processing of words derived by means of a suffix;

Fig. 61 is a block diagram of the full details obtained by merging Figs. 58 to 60;

Fig. 62 is a block diagram of further details for a complete unregistered word processing section in Fig. 61;

FIG. 63 is a block diagram for explaining an embodiment of an automatic transmission device in which the invention is applied;

64 to 90, a ninth embodiment of a speech analyzer which is Fig. According to the invention applied in a translation device English-Japanese, and show

64 is a functional block diagram of the entire structure;

Which summarized the function of detecting the structural arrangement of Fig 65 is a functional block diagram of an inputted English sentence as a block.

FIG. 66 is a flowchart showing an example of a flow for the collective arrangement of a block with respect to the input sentence;

67 is a flowchart showing the details of a word processing in the processing flow.

FIG. 68 is an explanatory view of an example of the dictionary information for English words or phrases that are stored in a word dictionary;

Figure 69 is an explanatory view of an example of table data for the block-Anfangszustand- final state and for a purpose and role evaluation state, which is stored in an analysis file.

FIG. 70 is an explanatory view of an example of a collective arrangement for a structure;

Fig. 71 is an explanatory view of an example of a collective arrangement for a block;

Figure 72 is an explanatory view for an example of English and word information, which are collectively disposed in a block.

Fig. 73 is a flowchart showing an example of an analysis processing which has been performed in a corresponding analyzing section;

Fig. 74 is an explanatory diagram similar to that in Fig. 68, showing an example of access to a word-phrase dictionary in the case that this embodiment has an identical case evaluation function;

Fig. 75 is an explanatory diagram similar to that in Fig. 69, showing an example of the start and end states of a block and a table of block preparation information for a case that this embodiment has an identical case evaluation function;

Fig. 76 is a functional block diagram similar to that in Fig. 64 and shows the whole construction of a modification of this embodiment;

Fig. 77 is a functional block diagram similar to that in Fig. 76 to summarize the function of a grammatical analysis of (let) information with respect to the modified embodiment shown in Fig. 76;

Fig. 78 is an explanatory diagram of an example of dictionary information containing the (let) information for English words and phrases stored in the word memory in the modified embodiment;

Figs. 79 and 80 are explanatory diagrams similar to those in Fig. 72 and showing an example of block and word information, an English sentence containing (let) information being collectively arranged in a block;

Fig. 81 is a flowchart similar to that in Fig. 73 and shows an example of a flow for a collective arrangement of the (let) information with respect to the input English sentence;

Fig. 82 is a flowchart similar to that in Fig. 73 and shows an example of the analysis processing which contains (let) information which has been performed in the analysis section in the modified embodiment;

FIG. 83A and 83B flow charts illustrating an example of the flow to the (let) information for the inputted English sentence grammar to analyze;

Fig. 84 is a flow chart of an example of the flow of the grammatical analysis for hyphenated words provided for the entered English sentence,

Fig. 85 is an explanatory diagram similar to that in Fig. 72, showing an example of the block and word information collectively arranged for the input English sentence containing a hyphen word in a block;

Fig. 86 is a functional block diagram similar to that in Fig. 64 and showing the entire structure of another modified embodiment;

Fig. 87 is a functional block diagram similar to that in Fig. 65 in which a function of a morphological analysis of the supplementary question in the input English sentence is collectively arranged in the modified embodiment shown in Fig. 86;

Figs. 88 and 89 are explanatory diagrams similar to that in Fig. 72 and show an example of collectively arranged block and word information for an English sentence containing a supplementary question in a block, and

FIG. 90A and 90B are flowcharts of an example of an analysis result of a supplementary question for the inputted English sentence.

The first embodiment of the invention will now be described. In Fig. 1, the overall structure of the first embodiment is shown in which a speech analyzer is used according to the invention in an automatic transmission device English-Japanese. Of course, the invention can be applied equally effectively not only to an automatic translation device for translating English to Japanese, but also to any language analyzer in which the sentences of an input language are mainly analyzed to translate one language into another.

The embodiment in FIG. 1 has an input section 110 , via which an English text 1012 , which is to be translated into Japanese, is input. The input section 1010 may include , for example, a keyboard with character keys such as alphanumeric keys or function keys, an optical character reader (OCR) for reading the English text recorded on paper, and / or a file memory for reading the English text which is written on a Storage medium such as a magnetic disk is recorded. The English text entered in the input section 1010 is read into a pre-editing section 1014 in which a pre-treatment for the translation is performed. This section mainly deals with sentence recognition and processing of unknown words. This then acts as part of a morphological analysis.

The pre-edited English data is transmitted to a morphological analysis section 1016 along with information contained in the pre-edited. Section 1016 divides the sentence with reference to a word dictionary 1018 , analyzes the morphemes of the English sentence, performs various types of arrangement such as processing for unknown words, expression for time, expression for numbers, etc., and processing for the entire sentence through like an additional survey and an identical case detection. The rule for the morphological analysis is stored in a rule file 1036 .

The English data after a morpheme analysis, together with the dictionary information obtained by the morphological analysis, are transmitted to a so-called parsing section I 1020 . (Here, parsing is understood to mean a grammatical analysis or an automatic syntax analysis). Section I 1020 is a functional section which performs an analysis for the surface structure of an English sentence by applying a grammar rule to the English data, and all structural possibilities are then found.

The English data after the analysis in section I 1020 together with its analysis information is fed to an analysis section II 1022 . In this section, a solution is selected by applying a structural description based on the result of the analysis with regard to the surface layer by section I. An acceptable "parsing tree" for the English sentence is thus prepared to form its structure. These parsing rules are also stored in the parsing rules file 1036 .

The English data is transferred to a structure transformation section 1034 after analysis as data for a "parsing tree". In section 1034 , a corresponding Japanese sentence tree is prepared from a structural tree, ie an intermediate structure of the English sentence, and is converted into a structure on which the Japanese is based, from which Japanese can then be easily translated.

The structural tree data showing the structure underlying Japanese, which has been subjected to structural conversion in this way, is supplied to a translation section 1026 , in which the translated sentence is formed. This is a functional section for creating a Japanese sentence from the tree structure of the Japanese structure tree.

The Japanese sentence data translated in this way, ie translated sentence data, is then delivered to a redesigning section 1030 . Section 1030 modifies the translated sentence data with reference to dictionary 10128 , using information that has been used in the translation to complete a more natural Japanese sentence. The Japanese sentence data is then transferred to an output section 1032 and then output as the translated Japanese sentence 1034 from the output section 1032 . The output section 1032 has, for example, a printer, a display and / or a file storage device, such as a magnetic disk.

The flow for the subsequent translation operations are controlled by a control section 1038 which regulates control for the entire facility. The word dictionary 1018 stores dictionary data for the words of the English and Japanese languages, specifying not only the vocabulary but also various information such as a linkage relationship, meanings, plural and singular forms, language parts, etc. in this embodiment will. File 1036 also stores the rule data for morphological and syntactic analysis.

The control section 1038 is connected to an operation display section 1040 . Section 1040 has operator buttons which give various commands from an operator to the device according to the invention, such as translation command buttons or cursor buttons, a display or a display device which visually displays an inputted English text, a Japanese sentence as a result of the translation, intermediate data such as displays dictionary information, various commands to the operator, etc. Most of the operation display functions are designed to be included in a keyboard when disposed on the input section 110 or in a display if provided on the output section 1032 .

In Fig. 1, the detailed structure of the morphological analysis section 1016 is shown. Section 1016 has an input unit 1100 , namely a keyboard for the input section 1010 and an input interface 1104 , which represents an interface with the file 1102 for inputted templates. The input interface 1104 is provided with a buffer for input character strings, which are input together with the data for English character strings in the form of code data, for example ASCII from the input unit 1100 or from the file 1102 , and temporarily stores the character series data. The character string entered may be the one that was pre-edited in section 1014 .

As shown in FIG. 1, the morphological analysis section 1016 has a processing section 1106 , a dictionary reference section 1108 , a contradiction removal processing section 1110 and a control section 1112 . The processing section 1106 is a parsing function section for performing the morphological analysis and has a buffer for queried dictionary information, ie a dictionary information preservation table 1110 (see FIG. 9). The morphological analysis is performed by properly commanding the dictionary to be queried from the top of the input string corresponding to the retrieve key string and by dictionary information obtained from the dictionary query section 1108 accordingly. is stored in the buffer 1120 for queried dictionary information, and that the processing of a preferred degree corresponding to the highest preferred flag is performed, as will be described later.

The dictionary query section 1108 is a functional section for extracting the dictionary information by querying the word dictionary 1018 based on the query key string that has been commanded by the processing section 1106 and then transmitting it to the processing section 1106 .

The word dictionary 1018 stores grammar information, such as a speech portion and inflection for access to each of the words, and a highest preference flag, as shown in FIG. 3 for the example of the access information. The dictionary is provided as a dictionary file with a highest preferred flag. "The highest preferred flag" is a flag that indicates the amount of coupling between words contained in a compound word or phrase that represents dictionary entry, in which "0" indicates weak coupling or no coupling, while "1 "indicates a strong coupling. In this case, language usage is evaluated as a phrase for a compound word or a phrase that is judged to have a strong coupling; on the other hand, the possibility of use in the form of individual words is also considered in parallel.

As illustrated in FIG. 3, each of the additions to word vocabulary 1018 is arranged for the compound word, phrase, and individual words that make up it, with no difference between the individual words and the compound word or phrase is. Furthermore, each diffraction form represents an access. If there are a number of diffraction forms, they are each registered as different inputs. The type of diffraction is shown in the diffraction section. The situation is similar to the language part in which registration for a number of language parts is permitted, and language part information is included for each of them. As further information, a predictability or a non-predictability for a noun, a transitive or intrasitive verb or a translated word etc. is registered.

For example, "get" is an infinitive form of a verb, and the highest preferred flag is "0". The phrase "get up" is a phrase for an infinitive form, and the highest preferred flag is "1". Furthermore, a preposition group "up to" has the highest reference flag "1", but a phrase such as "white house" as the compound word has the highest preference flag of "0", and consequently the latter shows that the degree of coupling between the words is low. In Fig. 3, the symbol indicates a space.

In this way, the dictionary information that was queried in the query section 1108 contains the highest preferred flag. If "1" is set for the highest preferred flag for identical series of characters or overlapping series of characters, such a contradiction must be eliminated. In section 1110 , the elimination of the objection is carried out and the subsequent processing, whereby reference is made to the highest preferred flag, which is stored in the file 1036 , with regard to the objection-elimination rule.

The objection-elimination provision is given in the present Embodiment in the following order (1) to (3) applied, whereby a preferred selection is carried out becomes.

• (1) phrase or word whose language part is a verb;
• (2) Compound word, phrase, or word with many Word components;
• (3) Compound word, phrase or word contained in the front part of the sentence.

The use for the word thus selected, that is, the parsing unit, is represented as the active information on the dictionary information buffer 1120 in the processing section 1016 . The active information shows that the parsing unit is valid or effective if it is "1", while it shows that its possibility is not used if it is "0".

The control section 1112 is a functional section for regulating and controlling the operation and processing in each of the functional sections in the morphological analysis section 1016 . The section may be included in the control section 1038 from which control for the entire facility is performed. The result of the morphological analysis is transmitted to the parsing section I 1020 by means of an output interface 1114 . In the event that the result is not immediately transmitted to the parsing section I 1020 , it is stored once in the parsing input file 1116 and in the parsing dictionary information file 1118 .

While in this embodiment all words, compound words and phrases starting from the corresponding position of the dictionary reference unit are taken out in a morphological analysis, dictionary information obtained for individual words becomes the compound word or form the phrase that has been judged as a collective unit according to the highest preferred flag. That is, the amount of coupling between the words in the sentence is judged while referring to the highest reference flag for the dictionary information obtained in the morphological analysis. The compound words or phrases that have been judged to have strong coupling are then evaluated to be used as a phrase in the sentence; if not, the possibility of using it in the form of individual words in parallel is also considered. Such processing by the highest preferred flag is carried out by the sequence shown in FIG. 4. Data for the input character series is received by the input section 1110 ( 1200 ); the entered series of characters is provided with a highest preference flag ( 1201 ) by a dictionary reference unit for a query in the dictionary file 1018 ; the dictionary 1018 is then queried again accordingly ( 1203 ) and the end position of the sentence represented by the data is carried out for the input character string ( 1202 ); then the contradiction for the highest preferred flag is resolved to 1204 and the result of the morphological analysis is given to parsing section I ( 1205 ).

In the input processing ( 1200 ), the data are first read from the file 1102 or an input unit 1100 into the buffer for input character strings of the input interface 1104 (see FIG. 5: 1210 ). For example, the data for the entered string of characters is entered in the form of ASCII; when data in the file is completely read out (for example, when the symbol EOF is read out), the processing section 1106 writes a NULL code in the buffer for the input character string as the closing position.

The processing section 1106 then reforms the input character string ( 1211 ). For example, if two or more characters belonging to a character corresponding to a space continue, they will be corrected into a single space. The character corresponding to a space contains spaces (which are represented by the symbol) tabs, line return (which is represented by the symbol) etc. These characters corresponding to a space between the upper and the first appearing characters, and not the character corresponding to a space in the buffer for the entered string is removed.

For example, the entered string or arrangement

as shown in Fig. 6, in

reshaped. The position of the symbol "ZERO" shows the end or end position of the memory at.

The dictionary reference boundaries used for the extracted dictionary reference unit processing 1201 are replaced by an alphabetical character, a numeric character, an apostrophe and other characters other than hyphen and paragraph, and excluding an apostrophe that follows spaces. arranged. Processing section 1106 has an upper pointer for a dictionary reference, which is set first at the top of the buffer.

The query section 1108 queries the dictionary file 1018 which has the highest preferred flag, using the character string from the character indicated by the upper pointer to the character preceding the next delimitation as the query key character string. The dictionary access and the query key string are compared; if both are identical, the dictionary information is admitted ( 1203 ). The order is judged if the entire string of characters of the access matches at least a part of the string starting from the top and if the part immediately after that part is a dictionary reference delimitation, an apostrophe or a paragraph. For example, as shown in Fig. 7, if the upper pointer indicates the upper character "g" in the query key string, it is true

the dictionary entry herewith match.

The retrieved dictionary information is stored in the buffer 1120 of the processing section 1106 . The matching start position and the end position of the character series are then stored together with the reading. This properly specifies the position of the characters in the input buffer from the top. An accumulation area for the active information is arranged in the buffer 1120 for queried dictionary information, which is information indicating whether or not the queried dictionary information is effective for the subsequent processing, in which step everything is set to "1".

The upper pointer is then updated with respect to the respective dictionary reference and is placed on the character immediately after the delimitation that appears closest to the existing upper pointer, which shows the row of characters from left to right. The dictionary reference is then carried out. In the above example, the character at the top of the dictionary reference is displayed first as "I" for "I", then as "w" for "will" and then for "g" for "get". When the upper pointer passes through the NULL code, it is judged whether that is the end position ( 1202 ).

Fig. 9 shows an example of the dictionary information which has been found in this way for the example of the English character string described above.

The contradiction clearing processing 1204 , which has been performed by means of the corresponding section 1110 , will now be described in connection with FIGS. 8A to 8D. The flowchart shown in Figs. 8A and 8B represents the processing in the case where the positions for the words where the highest reference flags are set overlap, while the flowchart in Figs. 8C and 8D shows the processing for parsing - Unit, ie removing elements with the highest preferred flag, ie processing to make the active information "0". In these flowcharts, the pointer "≦" represents a substitution, the character "→" a pointer and "P → x" the content of x , which is taken possession of by the input of the pointer p .

First, a set of words, each with the highest reference flag "1", and the positions in the sentence that overlap each other are determined (steps 1220 to 1223 ). Then the elimination rule regarding the highest reference flag is applied to each of the rates found and those at which it is effective are selected (steps 1224 to 1235 ).

In the embodiment described above, this is highest preferred flag "1" for "get up" at the start position "8" and "13" at the end position and "up to" at the start position "12" and the ending position "16" for the string

as shown in Fig. 9, and the positions for the characters then overlap each other. Then, the rule (1) mentioned above is applied first and it is judged whether it is a verb or not, referring to the speech part of the preservation pointer psave and the speech part of the pointer p ( 1224 ). Since it corresponds to a verb in this example, the combination "get up" is selected.

If rule (81) is not satisfied, rule (2) is applied ( 1228 ) and the length (lens) for the character arrangement , which relates to the input of the preservation pointer psave , and the length (len) for the character arrangement , which relates to the input of the pointer p is compared with each other. If rule (2) is also not satisfied, then rule (3) is applied ( 1229 ) and the position start, which relates to the start position for the preservation pointer psave , and the position start, which relates to the start position of the pointer p , are compared. Then, if one of the rules (1) to (3) is sufficient, if they are applied in this order, the active information is "not sufficient", ie an ineffective entry is made "NULL" ( 1232 ). Otherwise, while the active information is different, that is, an effective input is left because it is "1" ( 1231 ). Such application of the contrast elimination rule is performed sequentially, with the pointer p progressing ( 1234, 1235 ) to the end position for each of the inputs, and the active information is made only for the effective input "1". The state of the aforementioned example is shown in Fig. 10. For example, the active information "0" is made for the input "up to".

This overlap, even partial for the position with the combination in which both the active information and the highest preferred flag are "1", are found ( 1236 to 1241 ) and their active information is made "0" ( 1242, 1249 ). The application of such a counter-elimination rule is properly performed for each of the inputs with the pointer p incrementally ( 1243, 1248 ) to the end position and the active information of an ineffective input made "0". Consequently, the active information for "get" and for "up" is made "0", for example for the input "get up" ( FIG. 10). For the input "white", "white house" and "house", since all are the highest preferred flag "0", even if the positions overlap, their active information is given at "1".

If the processing has been performed until just before the end position (NULL) in this way, the contents of the input buffer of the input interface 1104 and the buffer 1120 for retrieved dictionary information from the output interface through interface 1114 to the parsing section I delivered 1016 . The content of the buffer 1120 is only output for the input for which "1" is displayed as the active information. For example, the content of the input buffer can be written to the parsing input file 1116 , while the content of the information buffer 1120 can be written to the parsing dictionary information file 1118 . In this case, since both the active information and the highest preferred flag are given, the structure of the information file 1118 is identical to that of the information buffer 1120 . However, it can also be agreed that the active information and the highest preferred flag are not given.

The invention will now be described with reference to a second embodiment of a speech analyzer according to the invention shown in FIG. 11, which is applied to an automatic translation device for English-Japanese. This embodiment has an input section 2014 , on which data is input from an input device 2010 or a template file 2012 . The input device 2010 has, for example, a keyboard with character keys, such as alpha / numeric or function keys and an optical character reading device, in order to read an English text recorded on paper. The template file 2012 is a storage device in which the English text is recorded on a storage medium such as a magnetic disk.

The input section 2014 has a buffer 2014 a for an input character string and stores the input English sentence, which was entered by the input device 2010 or the template file 2012 , in the buffer 2014 a . The input section 2014 reads out the input sentence, which is stored in the buffer 2014 a , and delivers it to a processing section 2016 .

The processing section 2016 is a functional section that performs the morphological analysis for the input sentence output from the input section 2014 by querying a dictionary file. The processing section 2016 includes a dictionary information preservation table 2016 a and stores the information which is obtained by querying a dictionary file 2022 or a basic unit dictionary file 2026, which will be described later, in the table 2016 a.

The processing section 2016 queries a query key string as a unit in the event that the dictionary is queried from the string representing the input sentence that has been input from the input section 2014 . The query string is properly queried based on the first character of the string representing the entered sentence according to a predetermined query rule. For example, the input sentence is properly split from the top of the sentence by delimitations such as spaces, commas, etc., and the split strings are used as the query key string. In this case, character strings which express units such as m, k, m / s are formed as query key character strings. Processing section 216 sends the query key string that has been retrieved from the string representing the entered sentence to dictionary query section 2020 .

Section 202 queries a dictionary file 2022 based on the query strings that have been issued by processing section 2016 . In the dictionary file 2022 , input and grammar information, such as a speech part, is stored as shown in FIG. 12. When there is an input in the dictionary file 2022 , the dictionary query section 2020 exposes the speech part information, etc. of that input and delivers it to the processing section 2016 . If there is no input to the dictionary file 2022 as a result of the query of the file 2022 , the dictionary query section 2022 passes this situation on to the processing section 2016 .

The processing section 2016 stores the speech part information etc., which has been retrieved from the section 2020 , in a dictionary information preservation table 216 a. If there is no input to the query string in the file 2022 , the processing section 2016 delivers the query string to a device recognition section 2024 .

Section 2024 queries a base dictionary file 2026 based on the query string that has been issued by processing section 2016 . The base unit inputs are stored in file 2026 , as shown in FIG. 13. If the basic unit input is present in the file 2026 , the unit recognizing section 2024 reads out the basic unit input. If there is no entry in the file 2026 , the query key string is divided into a number of strings, which will be described later, about the file 2026 is queried several times. Then, if there are basic unit inputs when the file 2026 is queried several times, a number of unit information are obtained from the basic unit inputs. If there is no basic unit entry in one of the plurality of queries, information is obtained which indicates that it is not registered in the dictionary.

The section 2024 outputs the basic unit input, composite unit information, and information indicating that the word is not registered in the memory to the processing section 2016 . This ( 2016 ) stores this information, which has been entered by section 2024 , in table 2016 a . The table 2016 stores to preserve the input for the query key string and grammar information, such as a language part, obtained by querying the file 2022 or the file 2026 for the query key string. After the data has been stored in table 2016 a , the processing section 2016 delivers this data together with the input sentence to the output interface 2018 . The interface 2018 outputs the input set and the data for the morphological analysis, which has been output by the processing section 2016 , to an output unit 2030 , such as a printer or a display unit, or to a storage file 2032 , such as a magnetic disk.

On the other hand, it is also possible to input the input sentence and the data of the morphological analysis, which has been output by the processing section 2016 , directly into a parsing device (not shown) in order to carry out a syntax analysis, ie the parsing for the input sentence in the Parsing facility and also to prepare a translated sentence based on the syntax analysis. The control section 2028 serves to control the functioning of each of the functional sections in the device according to the invention and can advantageously be designed in the form of a microprocessor.

The operation of the device according to the invention will now be explained with reference to the flow chart shown in FIG. 15. First, the entered English sentence is read from the input unit 2010 or from a template file 2012 into the input section 2014 ( 2100 ). The sentence read into the input section 2014 is stored in the buffer 2014 a for an input character string. The record in the buffer 2014 a is then read out and delivered to the processing section 2016 . When the sentence is entered in the processing section 2016 , the dictionary interrogator is cut out ( 2102 ). That is, the string representing the input sentence is divided into a query key string as the units for querying the dictionary file 2022 or the basic unit dictionary file 2026 successively, starting from the top of the string, according to a predetermined rule. It is then judged whether or not the split query string exists ( 2104 ); if it is present, the query string is delivered to the dictionary query section 2020 .

If the query string is submitted to section 2020 , this section ( 2020 ) queries file 2022 for the query key string ( 2106 ). It is then judged whether or not this character string is present in the input of the dictionary file 2022 , as shown in Fig. 12; if the input is present, the grammar information such as a speech part stored in the file 2022 is read out, and the read out data is sent to the processing section 2016 and stored in the table 2016 a ( 2110 ). The process then returns to step 2102 and the dictionary interrogator is turned off again.

If there is no input to file 2022 , section 2020 sends the query key string back to processing section 2016 . Section 2016 then delivers the query key string to the recognition section 2024 in which the unit recognition is performed ( 2112 ).

In the event that the query string from section 2020 has common words such as nouns and verbs, since they are mostly the inputs to dictionary file 2022 , grammar information, such as a language part, is read from file 2022 , and the data is given to the processing section 2016 and recorded in the table 2016 a . As described above, inputs are made for common words such as nouns and verbs, but no inputs are made for the string that expresses the units in file 2022 . Thus, in the event that the query key string is a string that expresses a unit such as kg or m / s since it is not an entry in the file 2022 , the process proceeds to step 2112 for a unit.

The unit recognition operation at step 2112 is explained with reference to FIG. 16. When the query key string for which there is no input to the file 2022 when queried is passed from the processing section 2016 to the section 2024 , the pointer P at the character at the top of the query key string in the section 2024 is set ( 2200 ).

Section 2024 then queries the base dictionary file 2026 for the character string starting from the character on which the pointer P is set ( 2201 ).

In this query, it is then judged whether or not the basic unit for which the input is available in the file 2026 appears as a complete character string in the character string starting from the character at which the pointer P is set, and whether it is started from the character on which the pointer P is set or not. Namely, a query is made as to whether or not a series of characters comprising one or a number of characters, starting with the character on which the pointer is set, matches any of the basic units for which the input is present in the file 2026 . For example, in the event that the characters at which the pointer P is set are k, m, s, etc., there are inputs in the file 2026 for those individual characters that begin at the character at which the pointer P is is set as shown in Fig. 13.

The unit recognizing section 2024 judges whether or not the inputs in the file 2026 are present as a result of the query on the file 2026 ( 2204 ). If the input is present, the pointer P advances by the length of the recognized basic unit ( 2208 ). Thus, in the event that the base unit is k, m, s, etc., the pointer P is advanced by one character and then set on the next character in the query key string.

The section 2024 then judges whether or not the string starting from the character on which the pointer P is set remains ( 2208 ). In the event that such a character string is still present, the process goes back to step 2202 , in which the file 2026 is queried again for the character string which starts from the character to which the pointer P is set. Then the section judges whether or not the input is present in the basic unit due to the query of the file 2026 ; if the input is present, the pointer P advances by the length of the recognized basic unit.

At step 2208, if the series of characters starting from the character for which the pointer P is set is no longer available, the retrieval at the file 2026 has ended, ie the recognition for the composite unit has been successful.

For example, in the event that the fetch key string issued to section 2024 is km / s, which is a unit, the input to file 2026 is not present because km / s is a complicated unit in and of itself. The pointer P is then first set to k ( 2200 ) and k is retrieved from the basic unit dictionary file 2026 to confirm the presence of the input ( 2202 ). Then the pointer P is set to m ( 2206 ) and m is retrieved from the file 2026 ( 2202 ) to confirm the presence of the input in the same way. Next, since the unit recognizing section 2024 considers a slash (/), solid circle (○), etc. as part of a unit, the pointer P is set to s, skipping "/" in km / s ( 2206 ) . Then s is retrieved from file 2026 to confirm the presence of the input in the same manner ( 2202 ). Since this was present for each of the units k, m and s when retrieved in the file 2026 , it is now judged that km / s is a string that expresses a unit. In this way, in the event that inputs in file 2026 are present for all of the characters that make up the retrieval key string, or in the event that inputs in file 2026 are available for all of the characters other than symbols such as slash and solid Circle, etc., which are already considered part of the unit, now judge that the retrieval key string is a string that expresses the unit.

If the unit recognizing section 2024 fetching finishes in the file in 2026 and has reached the goal to recognize the composite unit, it gives the unit information thus obtained to the processing section 2016 on what then is stored in the table 2016 a ( 2210 ). The unit recognition is then ended.

If, at step 2204, there is no input to file 2026 as a result of querying this file 2026 for the string starting from the character on which pointer P is set, it means that the string is not as a basic unit or a composite unit can be recognized. Therefore, the section 2024 outputs the information indicating that the string is a word that is not registered in the dictionary, that is, sends information indicating that the word does not express the unit back to the processing section 2016 , which is shown in FIG Table 2016 is a processing section 2016 of the conserved, which then causes the detection unit is completed.

When the unit recognition ( 2112 ) has ended, the flowchart of FIG. 15 goes back to step 2101 , and the section of the dictionary reference unit is performed again by the processing section 2016 . After cutting out the reference unit, section 2016 judges whether or not the unit cut out still exists ( 2104 ). If the cut-out unit, i.e. the retrieval key string is no longer available, it outputs the information stored in table 2016 a to the output unit using the output interface 2018 ( 2114 ). The syntax analysis or the so-called parsing for the entered sentence is thus completed. As described above with respect to this embodiment, the entered English sentence is divided into retrieval key strings and first retrieved from an ordinary dictionary file 2022 ; if there is no input in file 2022, device detection is performed. In unit recognition, the fetch key string is split and indicated by the pointer P , and the base unit dictionary file 2026 is fetched at the respective split string. What is recorded in file 2026 or what is composed of a series of rows recorded in file 2026 is then judged whether it is rows of characters which express units.

As a result, since it is possible to perform unit recognition even for the string expressing a complicated unit by combining basic units stored in the file 2026 , so-called parsing can be performed, which corresponds to versatile unit expression. In addition, since it is only necessary that the file 2026 basic units such as k, m, s,. . . etc. stores and complicated units composed of them, such as km, ○ km / s, etc. need not be stored, the capacity of the dictionary file can be reduced.

The overall structure of a third embodiment of the invention in which the speech analyzer according to the invention is applied to an automatic English-Japanese translation device will be described with reference to FIG. 18.

This embodiment has an input section 3010 via which an English text 3012 to be translated into Japanese is input. The input section 3010 may include , for example, a keyboard with character keys such as alphanumeric or function keys, an optical character reader (OCR) for reading the English text recorded on paper, and / or a file storage device for reading the English text written on the Storage medium such as a magnetic disk is recorded. The English text input from the section 3010 is read into the pre-editing section 3014 , in which a pre-treatment for the translation is performed. In this case, recognition for the sentence and processing of unknown words are mainly performed. This then acts as part of a morphological analysis. After the pre-editing, the English data is transmitted to a morphological analysis section 3016 together with the information obtained during the pre-editing. Section 3016 analyzes the morphemes of the English sentence while being divided by retrieving a word dictionary 3018 , performs various arrangements, such as processing for unknown words, for a noun, for a time expression, for numbers, etc., and performs the processing for the entire sentence, such as an additional question (tag question) and recognition of an addition. The morphological analysis rules are stored in a corresponding rule file 3036 .

After the morphological analysis, the English data are stored in a parsing section I 3020 together with the dictionary information obtained in the morphological analysis. Section I 3020 is a functional section which analyzes the surface layer structure for the sentence by applying a grammar rule to the English data, and it finds out all the possibilities with regard to the structure.

The English data which have been subjected to the morphological analysis in section I 3020 are transmitted together with the morphologically analyzed information to a parsing section II 3022 . In this section, a solution is selected by applying a structural description based on the result of the morphological analysis (parsing) with respect to the surface layer structure by the syntactic analysis I, thereby preparing a plausible parsing tree for the English sentence to shape its structure. These parsing or analysis rules are stored in the rule data 3036 .

The English data which has been subjected to the corresponding analysis is transmitted to a structure converting section 3024 as the data for the so-called parsing tree. In section 3024 , a structure tree of a corresponding Japanese sentence is prepared from the structure tree, which is an intermediate structure of the English sentence, in order to convert the Japanese sentence into a structure underlying the Japanese, from which a Japanese sentence can then be easily translated. The structure tree data indicating the underlying Japanese structure which has been transformed in this way is supplied to a translation forming section 3026 , in which the translated sentence is then formed. This is a functional section to build a Japanese sentence from the structure of the Japanese sentence structure tree.

The Japanese sentence data prepared as a translated sentence, that is, translated data is supplied to a pre-editing section 3030 . Section 3030 modifies the translation data that is retrieved from a dictionary 3018 using information that has been used in the translation to complete a more natural Japanese sentence. The data for the Japanese sentence is transferred to an output section 3032 and output from it as a translated Japanese sentence 3034 . The output section 3032 includes, for example, a printer, a display, and / or a file storage device such as a magnetic disk.

The flow of a series of translation operations is controlled by a control section 3038 which regulates control for the entire facility. The word dictionary 3018 in this embodiment stores dictionary data for English and Japanese words, in which various information is described, such as a connecting relationship, that is, concurrent relationships, meanings, plural or singular forms, parts of speech, etc., in addition to the vocabulary . File 3036 also stores control data for morphological and syntactic analysis.

The control section 3038 is connected to an operation display section 3040 . The section 3014 has, for example, operation keys such as a translation display key or a cursor key to give various information from the operator to the corresponding device, has a display or a display device that visually shows the inputted English text, the Japanese text as a result a translation, intermediate data such as dictionary information etc. and various displays for the operator. Many of these functions indicative of an operation may be configured to be included in a keyboard if it is located in the input section 310 , or on a display when it is located in the output section 3032 .

FIG. 17 shows an example of a detailed structure for processing numbers in the morphological analysis section 3016 . Section 3016 of course also has another functional section, but these parts which relate directly to the understanding of the invention are also shown here. The morphological analysis is performed by commanding the dictionary query from the upper end of the input string one by one in accordance with the retrieval key string, and processing the dictionary information obtained from the dictionary query section 3104 according to a numerical flag what will be described later.

Section 3016 has an input processing section 3100 for receiving and processing the data for the character string input from preprocessing section 3014 . Section 3100 is provided with an input string buffer which receives input of English string data in the form of code data such as ASCII data and temporarily stores the string data.

The input string data temporarily stored in section 3100 is provided to a unit splitting section 3102 which splits the input string data into dictionary query units such as words. Section 3102 is a functional section to distinguish the dictionary reference units that represent a retrieval key string when retrieving dictionary 3018 in section 3104 . The dictionary reference boundaries that have been used for dividing the dictionary reference unit are arranged at a specific position of an English character, a numeric character, an apostrophe, a character other than a hyphen and a paragraph, and at an apostrophe that precedes one Space follows. This is stored in a limit table 3108 in 99999 00070 552 001000280000000200012000285919988800040 0002003733674 00004 99880 and this is referred to in section 3102 when the dictionary reference unit is split. The word dictionary 3018 contains in particular information for querying the divided units. As illustrated by the example of the input information in Fig. 24, this is done for the input of each of the dictionary reference units, e.g. B. for words, grammar information such as a sentence and to distinguish a display which indicates that a word represents a number, ie a numerical flag and a numerical value information which indicates the numerical value for a word indicating the number.

As shown in Fig. 17, both singular and plural forms are described along with each of the inputs in the word dictionary 3018 , and each of them represents an input. The numeric flag indicates that a word means a number if for this "1" is set. As further information, for example, a countability and non-countability for a noun, an identification for transitive or intransitive verbs, for translated words, etc. are registered. Since, for example, when referring to "thousand", this is a noun that represents a number for this the numerical flag "1" and the numerical value is "1000". Since with reference to "thread" this is a noun, but not a noun indicating a number, ie not a number, the numerical flag is registered as "0".

The recognition of the number is therefore carried out by the numerical flag in the event that it is a word which is registered in the dictionary 3018, for example, as "one" or as "thousand". Even the unregistered words, such as a sequence of digits such as "123", two sets of digit sequences with a "dot" between them, namely a small number such as "10.2" and also the sequence of numerical characters with a comma between them , such as "1,000,000" are also recognized as numbers. In the present specification, the term "numeric character" usually reads not only Arabic numerals, but also a full numeric expression such as "thirteen".

As shown in Fig. 28, the dictionary 1318 has a currency symbol table 3018 a in which various currency symbols are registered, a notation symbol table 3018 b in which notation symbols ","".""(Space)" are registered, and a decimal point table 3018 c , in which decimal points "."",", etc. are registered. The tables for the notation symbols or the decimal points are arranged because "," is used for the notalion symbol, or "." is used for the decimal point in Japanese or English, while a space or "period" is mainly used for the notalion symbol and "," for the decimal point in other European languages such as German or French; consequently, the use of symbols differs between languages to be processed.

The dictionary retrieving section 4104 is a functional section which divides the dictionary information by retrieving the word dictionary 3018 based on the retrieving string input from the input dividing section 3102 and transmits it to the processing sections 3110, 3112 and 3116 .

A compilation for the sequence of numerical Character is processed through the following two carried out. If words are recognized as a number, like as described above when the next dictionary- Reference unit is referred to and it also as a Number is recognized, they are first arranged together to make them put together in a single number. The surgery is going repeated as long as digits follow. For example "30 thousand" converted to "30000", and "1.5 million" will transformed into "1500000". Then if the numeric expression is continued, with "and" in between if all the digits to the left of "zero", each of which Digits of the numerical values correspond to what is indicated by the Pointer to the right of "and" is displayed with respect to the Meaning of the numerical expression "0", they are in  summarized a number. For example, "one hundred and thirty" summarized in "130", while "30 thousand and two hundred "is summarized in" 30200 ".

After the recognition as a number, the necessary local one becomes Analysis, the so-called parsing continued. At this processing is a series of so-called Parsing units identified by the morpheme actuation information for each of the parsing units are arranged together in a single parsing unit, which is based on a local analysis rule. For example are a currency symbol and a numerical value like "Y1,000" arranged collectively as "thousand yen", and one numerical value and a unit like "1.5 km" arranged.

These arrangements are performed in processing sections 3110 through 3122 . The processing section 3210 is a functional section for collectively arranging a number together with a currency symbol or a unit. The processing section 3112 is a functional section for performing numbering for the number. Further, the processing section 3114 is a functional section for processing numbers connected by a hyphen. The processing section 3116 is a functional section for processing successive numeric characters.

Regarding the number after the combination with a currency symbol or a unit, the combination between the currency symbol and the numerical number is performed into a single noun in the processing section 3118 . Further, the compilation between the unit and the numerical value into a proper name is performed in the processing section 3120 . Further, in the case of a number that has been subjected to numbering processing, in the case of the hyphenated number and for a consecutive number, processing for compilation with the preceding numerical value is performed in the processing section 3122 . The dictionary information for the input character string, which is completed by such processing, is stored in the ordered dictionary information buffer, that is, in the dictionary information preservation table 3124 . The results of the morphological analysis are transferred from table 3124 to parsing section I 3020 . Processing by the numerical flag is performed sequentially, as shown in Figs. 19A and 10B. The data for the input character string is received at the input processing section 3100 , where input processing is performed ( 3200 ). Then, the unit division section 3102 divides the input character string into dictionary reference units for retrieving the dictionary 3018 ( 3201 ). The dictionary retrieving section 3104 searches the dictionary 3018 accordingly ( 3203 ) and if there is a dictionary entry ( 3204 ) it checks the numerical flag ( 3205 ). If the numeric flag is not set because the word is not a number, the dictionary information is collected in table 3124 . If "1" is set for the numerical flag, the number in the processing section 3112 is numbered ( 3206 ) and the collective assembly 3207 with the previous numerical value is performed in section 3122 . When this processing is done for the end point of a sentence, which is indicated by the input character string data ( 3202 ), a collective composition ( 3209 ) between the currency symbol or the unit in sections 3118 and 3120 is performed, the result of the morphological analysis is then shown in submitted the syntactic analysis section I ( 3020 ) ( 3210 ).

If, as a result of the dictionary reference, there is no input at step 3204 and if the element is added with a hyphen, processing for a hyphened number ( 3213 ) is performed in section 3114 . If the first character is not a hyphenated character but a currency symbol ( 3214 ), the currency symbol alone is preserved in table 3124 ( 3216 ) and the currency symbol is deleted from the dictionary reference unit ( 3217 ). If the first character is not a currency symbol ( 3214 ), processing for the subsequent numeric characters ( 3215 ) is performed in processing section 3116 . The operation is carried out to the end point ( 3202 ).

The collective assembly ( 3209 ) with the currency symbol and the unit is performed in the processing section 3110 by the processing flow shown in FIG. 20. First, in initial processing ( 3220 ), the upper processing pointer is placed first on the top of the buffer. If the element indicated by the pointer is not a numerical value ( 3221 ), the pointer is gradually shifted ( 3226 ). In the event that the character is a numerical value, but has no preceding currency symbol and no preceding unit, the pointer is also gradually shifted further ( 3222, 3224 ). Processing is carried out to the end point of the dictionary reference unit ( 3227 ).

If the character is a numerical value ( 3222 ), the currency symbol and the numerical value are collectively combined into a single noun ( 3223 ). For example, the currency symbol and the numeric character "1,000" are combined to form a noun. Furthermore, if the preceding character is not a currency symbol and the subsequent character is a unit, the numerical value and the unit are collectively combined into a single noun ( 3225 ). For example, a numeric character and a unit "1.5 km" are collectively combined into a single noun. Processing is carried out to the end point of the dictionary retrieval unit ( 3227 ).

The processing of the hyphened number is performed in the processing section 3114 by the flowchart shown in Figs. 21A and 21B. First, the hyphenated dictionary reference unit is stored in the buffer during initial processing ( 3230 ). Furthermore, the numerical value "0" is preserved and the hyphen in the original dictionary reference unit is changed to a space. Then the dictionary retrieval unit is divided ( 3231 ) to perform retrieval in the dictionary ( 3235 ). If there is no input as a result of dictionary retrieval, that is, if the word is not registered in the dictionary ( 3236 ), the entire hyphenated dictionary reference unit is stored as a word not registered in the dictionary in table 3124 ( 3237 ).

If an input is received as a result of the dictionary retrieval ( 3236 ), it is checked whether the numerical flag therefor is "1" or not. If the numeric flag is not a "1", it means that the character is not a numeric character and the entire hyphenated reference unit is kept as a word not registered in the dictionary in table 3124 ( 3237 ).

If "1" is set on the numeric flag for the dictionary input, the section 3012 numbers the number based on the input data ( 3239 ). Then the numbered numerical value is added to a numerical value kept at this time ( 3240 ) and the addition result is preserved ( 3241 ). Thus, for example, "two" in "twenty-two" is added to "3020" of "twenty" immediately before to "3022". Processing is carried out up to the end point of the dictionary retrieval unit ( 3232 ).

If incrementally advanced to the end position, the flow is transferred to processing 3233 at step 3232 and the stored numerical value is made as a numerical value for the entire hyphenated dictionary reference unit. Then a collective composition 3207 for the numerical value is performed together with the previous numerical value. The following numerical character processing 3215 will now be explained with reference to FIGS. 22A and 22B, which is carried out in the processing section 3116 . In these flowcharts, the symbol "" means substitution. First, an initialization 3250 is carried out, in which the stored numerical value val - save is set to "0", the parameter "i" is set to "1" and the pointer p is set to the upper end of the character series of the dictionary reference unit.

It is then checked whether the character * p indicated by the pointer p is a numeric character ( 3251 ), a notation character ( 3252 ) or a decimal point ( 3251 ). If it is not one of them, the entire string is stored as the word not registered in the dictionary in table 3124 ( 3255 ). If it is a decimal point ( 3253 ), parameter (i) is multiplied by 10 ( 3254 ) and step ( 3258 ) is performed. At step 3258 , the numerical value num (* p) for the character * p is added to the stored numerical value val - save in order to prepare a new stored numerical value. The numerical value num (* p) is a value that considers the character (* p) as a numerical value.

At step 3251 or 3252, if the character is a numeric or a notation character, step 3257 is performed. At step 3257 , the stored numerical value val - save is multiplied by 10, to which the numerical value num (* p) for the character * p is added in order to prepare a new stored numerical value.

After this processing, the pointer is gradually advanced ( 3259 ) and the processing is repeated to the end point of the dictionary reference unit ( 3260 ). When the end position for the string is reached, the numerical value for the whole string is formed as a conserved numerical value ( 3261 ) and a collective compilation 3207 with the previous numerical value is carried out in section 3122 . The following numerical characters, e.g. B. "1,000.5" analyzed as a numerical value "1000.5".

The collective compilation 3207 with the previous numerical value is carried out in the section 3122 as follows. First, the dictionary table pointer is set to a previous location of the dictionary reference unit ( 3270 ). If nothing is present at this point, it means that the first digit in the preservation table indicates the numerical value and that the numerical value for the current dictionary reference unit is recorded in table 3124 ( 3284 ). The recorded location is the location closest to the location indicated by the pointer P.

If at step 3271 there is a word at the previous position when the input indicated by pointer p is not "and" ( 3272 ) and pointer p does not indicate the numerical value ( 3273 ), the numerical value for the current dictionary is Reference unit recorded in table 3124 which is closest to the current position indicated by pointer p ( 3284 ). In the example "To him two..."("To him two..." is "two" recently recorded as a numerical value "2".

If at step 3273 the pointer p indicates a numerical value, the numerical value p → v for the input indicated by the pointer p is multiplied by the numerical value v -now for the current dictionary reference unit by a new numerical value p → v for the to form the input indicated by pointer p ( 3274 ). In the case of "two thousand", for example, "2 × 1000 = 2000" is carried out in order to combine the entire "two thousand" into one printout. Then the end position for the current dictionary reference unit is set as the end position for the input of the pointer p, the p -end position ( 3282 ).

At step 3272, if the input indicated by pointer p is "and", pointer p before that is transmitted to the dictionary retriever ( 3275 ). If it is not at the end point ( 3276 ) and if it is a numerical value ( 3277 ), the numerical value v -now is transmitted to the current dictionary reference unit and rounded off to the most significant digit, which is then set as a value vl . If the numerical value v -now of the current reference unit is "8", "8.1", "98" or "11", for example, the value vl is "10", "10", "100" or "100".

It is then checked whether the excess obtained by dividing the numerical value p - v for the input indicated by the pointer p by vl , that is, mod (p - v, vl) is "0" or not. If it is not "0", pointer p is incremented ( 3283 ) and the numerical value for the current reference unit is recorded in table 3124 at a position which is closest to the position indicated by the current pointer p ( 3284 ). For example, in the case of "I and two", "two" is recently recorded as a numerical value "2".

If the excess at step 3279 is "0", the numerical value v -now for the current reference unit is added to the numerical value p → v for the input indicated by the pointer p by a new numerical value p → v for the through to form the pointer p indicated input ( 3280 ). In the case of "two thousand and two", for example, "two thousand" are already combined to "2000" at this level. Then it is added with "2" of "two" by the addition 3200 so as to collectively assemble the whole part into "2002". Then the information "and", which is indicated by the pointer p + 1, is deleted from the table 3124 ( 3281 ) and the process proceeds to step 3282 . Retrieval is now explained using an example. For example, if a dictionary retrieve the string "To him two thousand and twenty-two ..." (To him two thousand and twenty two... ") Is performed, as shown in Fig. 25, the dictionary input information is written into the table 3124 as shown in Fig. 26A. For" him "(" him ") For example, the start position is" 4 ", the end position is" 6 "and the speech part is a pronoun. For numerical processing, it is first judged for" two "that the numerical flag is" 1 "( 3205 ) and the numerical value therefor "2." Since the preceding character to "2" in this series of characters is not a numerical value, it is immediately stored in table 3124 ( 3206, 3208, 3284 ).

Then the pointer is incremented to proceed to processing "thousand". The numeric flag is "1" and the numerical value is "thousand" ( 3205, 3206 ). In addition, since the numerical value of the previous character is "2" ( 3207, 3273 ), the multiplication: 2 × 1000 is performed ( 3274 ), and its result is stored in the table 3124 (see Fig. 26B). For the next "and", the dictionary information is temporarily stored as it is in the table 3124 (see Fig. 26C).

The pointer is then advanced to process "twenty-two". Since the words are a hyphenated word that is not found in the dictionary ( 3212 ), "20 + 2 = 22" is performed in processing 3213 for the hyphenated numbers ( 3237, 3239 to 3241 ). Since the preceding word is "and" ( 3272 ) and the preceding numerical value is "2000" ( 3277 ), the numerical value "22" is rounded down to "100" at the most significant position ( 3278 ) and a division operation 3279 is carried out. Since the excess is "0", an addition "3280" between "2000" and "22" is performed. The information for "and" is removed from table 3124 ( 3282 ) and the result of the addition ( 2022 ) is stored as a numerical value in table 3124 , whereby "two thousand and twenty-two" is recognized as "2022". As a result, a collective composition with the previous numerical value has been performed ( 3207 ). Another example will now be presented. As shown in Fig. 27, so-called parsing for the entered string "it said $ 1,000.5 thousand was ..." (You said $ 1,000.5 thousand was...). "$ 1,000.5" was not registered in the 3018 dictionary. The first character is the currency symbol "$", which can be recognized as the currency symbol from the dictionary input. This is independently recorded in table 3124 ( 3214, 3216, Fig. 29A).

Then "1,0005." into a numerical value "1000.5" by the subsequent numerical character processing 3215 . Since the preceding character is the symbol "$" and not a numerical value, the numerical value itself is recorded ( 3280 to 3273, 3284, Fig. 29B).

The next word "thousand" is a number and its numerical value is "1000". Since the preceding character is a numerical value ( 3272, 3273 ), a calculation is made: "1000.5 × 1000 = 1000500" ( 3274 , Fig. 29C). After dictionary retrieval is finished, the content kept in table 3174 is checked one by one. Since the currency symbol "$" is immediately before the numerical value "1000500", both are put together collectively, and "$ 1000500" is formed as a single phrase ( 3209, 3221 to 3223 , Fig. 29D).

The fourth embodiment according to the invention will now be explained. Fig. 30 shows the fourth embodiment of the speech analyzer according to the invention, which is applied to an automatic translation device for English-Japanese.

This embodiment has an input processing section 4014 in which data is input from an input device 4012 . The input device 4012 has, for example, a keypad with character keys, such as alphanumeric and function keys, an optical character recorder for reading English text recorded on paper, and a reader, such as a magnetic disk.

The input processing section 4014 has a buffer 4014 a for an input character string and stores the English sentence entered by the device 4012 in the buffer 4014 a . The section 4014 reads an input sentence stored in the buffer 4014 a and delivers it to the unit division section 4016 . Section 4016 is a functional section that divides the dictionary reference unit of the sentence entered by section 4014 . A delimitation table 4018 stores delimitations, such as space and comma. Section 4016 stores delimitations from delimitation table 4018 and divides the input sentence from section 4014 into character strings as the units in the case of retrieving a reference memory 4020 by dividing the sentence where the deductions exist. The divided character strings are input to the dictionary retrieving section 4022 .

Section 4022 retrieves the reference dictionary 4020 for the input sentence, which is divided into reference units provided by section 4016 . The reference dictionary 4020 stores inputs for, for example, the English strings, parts of speech thereof, feature information, etc. as shown in FIG. 31. The dictionary 4020 additionally stores, for example, proper names, character strings shown in the figure, for another language part, verbs and adjectives. The record for the proper name as a language part in the figure means that this is applied in the processing for the registered proper name, which will be described later, but that no common grammatical proper names are expressed. Furthermore, the feature information indicates what is represented by the proper name, which cannot always be limited to one, since, as will be described later, a proper name represents a number of features depending on the application.

Section 4022 retrieves dictionary 4022 for the string divided into reference units, and if the string is a proper name, it is passed to proper name processing section 4024 , where processing for the proper name is performed, which will be described later. If it is not a proper name, this is given to the processing section 4036 and stored in its table 4036 a .

The proper name processing section 4024 has a processing section for the previous sentence end, a processing section 4028 for a previous proper name, a processing section 4030 for the proper name per se, a processing section 4032 for the previous proper name and the proper name per se and a section 4034 for a default or standard feature information. Section 4026 judges whether or not the string preceding the string retrieved by section 4022 and entered from section 4022 is at the end of the sentence if the previous string is at the end of the sentence. it converts the uppercase letter to a lower case at the beginning of the character string to be processed, passes it to the dictionary retrieving section 4022 , and causes section 4022 to retrieve the reference dictionary 4020 again. The character string that was not retrieved even by the second retrieval process is judged to be an unregistered proper name, which is then sent to the processing section 4036 and then stored in the table 4036 a . Furthermore, if the preceding character string is not at the end of the sentence, it is given in the processing section 4036 as a proper name whose feature information is unknown and is registered in the table 4036 a , as will be described later.

Section 4028 dissects the feature information for the previous string from section 4026 and delivers the result to the proper name processing section 4030 per se. The section 4030 checks the feature information of the proper name to be broken down, and as will be described later, when the feature information of either the proper name or the previous proper name is not registered, it decomposes the proper name and the previous proper name collectively by the registered information for the other of them and keep it in table 4036 a in section 4036 .

The previous proper name and the proper name processing section 4032 itself checks the part that is common to the characteristic name information for the proper name and for the previous proper name to be decomposed, decomposes these proper names with respect to the common part, outputs the result to the processing section 4036 , and stores him in his table 4036 a . Section 4036 creates feature information on a proper name after it is read from table 4036 a , sent to section 4022 via section 4016 and, as a result of querying dictionary 4020 , section 4022 has found that there is no feature information. Since it is effective that a proper name has different types of features depending on the use cases, all feature information is provided which must necessarily be considered. For example, "person, place, group, etc." intended. After the proper name is provided with the feature information, the section 4034 sends the data to the processing section 4036 and stores it in its table 4036 a .

The section 4036 with the table 4036 a stores the data from the section 4032 for the previous proper name and the proper name per se, the section 4034 and the retrieving section 4022 in the dictionary information preservation table 4036 a and then reads out and specifies the stored data parsing section 4038 . Section 4038 carries out the analysis for the input sentence after it has been subjected to a morpheme analysis and it has been read from table 4036 a .

The operation of this device will now be explained with reference to the flow chart in Fig. 32. First, an English sentence input from the device 4012 is read into the input processing section 4014 ( 4100 ). The read section 4014 in the set is read into the buffer 4014 a load, which is in the buffer 4014 a charged inputted sentence in the unit Aufteilabschnitt 4016th When the sentence is entered, section 4016 reads delimitation table 4018 to divide the dictionary reference units ( 4102 ). That is, to follow characters representing the input sentence are sequentially divided into fetch strings from the top of the strings as the unit for fetching the reference dictionary 4020 by dividing them into the parts where delimitations such as there is a space or a colon. The section judges whether the divided dictionary reference unit, that is, fetch strings are finished or not ( 4104 ), and if fetch strings still exist ( unfinished ), it passes the fetch string to section 4022 .

If a fetch string is provided to section 4022 , section 4022 fetches the reference dictionary 4022 for the fetch string ( 4106 ). The section judges whether the retrieval character string in the input of the reference dictionary is present 4020 or is not shown in Fig. 31 (4108), and if an input exists, it reads out the information stored in the dictionary 4020 part of speech information and assessed whether the retrieval string is a proper name or not ( 4110 ). If the retrieval string is not a proper name, the section 4022 delivers the data read from the dictionary 4020 to the processing section 4036 and records it in its table 4036 a ( 4112 ). When the data is stored in the table 4036 a , the input indicating that the data is stored in the section 4016 and the data for the retrieved string just saved are input from the processing section 4036 . Then, step 4102 is returned to and the dictionary reference unit is again divided into section 4016 .

If, in step 4110, the retrieval string is a proper name, section 4022 passes the data for the proper name read from dictionary 4020 , hereinafter referred to simply as proper name, to proper name processing section 4024 along with the data for the previous one polling data sequence which has been inputted from the table 4036 a in the portion 4036, by means of the section 4016 to the section 4022 decreases and the processing for the registered in the dictionary proper names is performed in the section 4024 (4124).

The processing for the proper name registered in the dictionary will now be explained with reference to the flow chart in FIG. 33. The data output from section 4022 to section 4024 is input to processing section 4028 for the previous proper name by processing section 4026 for the previous sentence end. Section 4026 has no function when processing the proper name registered in the dictionary. The processing section 4028 judges whether or not the input that precedes the proper name is a proper name not registered in the dictionary 4020 , that is, whether it is used in the processing for the proper name not registered in the dictionary, which will be described later, or not ( 2100 ). If it is an unregistered proper name, the processing section judges the whole part of the proper name and the previous unregistered proper name as a proper name having the characteristic information for the proper name ( 4202 ), and outputs the data to the processing section 4036 and stores it in its table 4036 a ( 4218 ).

Subsequently, if the section 4028 judges at step 4200 that the input before the proper name is not a registered proper name, then it judges whether the input before the proper name is a proper name that registers in the dictionary 4020 or not ( 4204 ). If the input just before the proper name is a registered proper name, it is judged whether or not the feature information for the previous proper name is known, that is, whether it is not registered in the dictionary 4022 ( 4206 ).

If the feature information for the previous proper name is unknown, the process proceeds to step 4202 to judge the entire portion of the proper name itself and the proper name preceding it as a proper name with feature information ( 4202 ); the processing section 4028 for the previous proper name then outputs the data to the processing section 4036 , in whose table 4036 a the output data are then recorded ( 4218 ).

If the processing section 4028 judges at step 4206 that the feature information for the previous proper name is not unknown, that is, that it is registered in the dictionary 4020 , the data is supplied from the section 4028 to the processing section 4030 . Section 4030 then judges whether or not the feature name information is unknown ( 4208 ). If the feature information for the proper name is unknown, the section 4030 judges the entire portion of the proper name per se and the immediately preceding proper name as a proper name that has feature information of the proper name ( 4210 ) and outputs the data to the processing section 4036 , in the table 4036 a the data are recorded.

If the processing section 4030 determines that the characteristic information of the proper name per se is not unknown, that is, that it is registered in the dictionary 4020 , the section 4030 outputs the data to the processing section 4032 for the previous proper name and the proper name itself. The processing section 3032 then judges whether there is any common feature in the feature information between the proper name per se and the proper preceding name ( 4212 ). If there is a common feature, it judges the entire portion for the proper name per se and for the immediately preceding proper name as a proper name that has the feature information of the common part ( 4214 ) and outputs the data to processing step 4036 , in which Table 4036 a the data is then recorded ( 4218 ).

If there is no common feature for the feature information between the proper name itself and the immediately preceding proper name, the processing section judges that the proper name is a proper name with the feature information retrieved from the memory 4020 that is different from the immediately preceding one Proper names distinguishes ( 4216 ), and delivers the data to the processing section 4036 , in whose table 4036 a the data are then recorded ( 4218 ). In Fig. 32, if there is no fetch string in the input of the reference dictionary 4020 at step 4108 , it is judged whether or not the first character of the fetch string is an uppercase letter, and if it is not an uppercase letter, section 4022 considers that Fetch string as an unregistered word, outputs it to the processing section 4036 so that it is recorded in its table 4036 a ( 4118 ).

If the first character is an uppercase letter, the data for the retrieval string along with the data for the previous retrieval string are passed from the section 4022 to the section for proper names, where the processing for the non-dictionary registered proper name is performed ( 4120 ).

The processing of proper names not registered in the dictionary will now be described with reference to FIG. 34. The data for the fetch string, along with the data for the previous fetch string, is provided to the processing section 4026 , which then judges whether or not the end of the previous input is a candidate for the end of the sentence ( 4300 ). The judgment as to whether or not it is a candidate for the end of the sentence is made by judging whether the end of the previous input is a candidate for the end of the sentence, such as a separate period (.), Etc. or not .

If the end of the previous input is a candidate for the end of the sentence, data is passed from processing section 4026 to processing section 4028 , which then considers the previous input as the end of the sentence ( 4302 ), the first character in the retrieval string converted to a small letter and passed it to section 4022 .

Section 4022 retrieves dictionary 4020 for the fetch string that has been converted to the lower case character ( 4304 ) and judges whether there is an input to reference dictionary 4020 or not ( 4306 ). If there is an input, section 4022 passes the data retrieved from dictionary 4020 to processing section 4036 and stores it in table 4036 a ( 4308 ). If there is no input, section 4022 reverts the first character in the fetch string to the uppercase character, passes it as an unregistered proper name to processing section 4036 , thereby recording it in its table 4036 a ( 4310 ).

At step 4300 , if the previous sentence end processing step 4026 judges that the end of the previous entry is not a candidate for the end of the sentence, the data is passed from section 4026 to section 4038 which then judges the previous entry so that it is not the end of the sentence ( 4312 ). The data from section 4028 is provided to section 4030 , which then views the fetch string as a proper name whose feature information is unknown ( 4314 ).

Section 4030 then returns the data to processing section 4028 , which then performs processing for the proper name registered in the dictionary ( 4316 ). The processing for the proper name registered in the dictionary is the same as that shown in Fig. 33.

If the reference unit at step 4104 is now at the end in FIG. 32, section 4022 emits a signal which indicates this to section 4034 , which then reads out the information recorded in table 4036 a in section 4036 and the proper name with the default -Feature information creates ( 4122 ).

The operation will now be described with reference to FIG. 35 to provide the proper name with the default feature information. In the section 4034 provided for this purpose, a pointer is first set to the top of the data in the table 4036 a ( 4400 ). That is, the pointer is placed on the input at the top of the input set, which are divided into inputs, each of which is provided with information by retrieval in the reference dictionary 4020 . It is then judged whether the input indicated by the pointer is a proper name or not ( 4402 ); if it is a proper name, it is judged whether or not the characteristic information of the proper name is unknown ( 4404 ). If it is not a proper name, the flowchart proceeds to step ( 4408 ) and the pointer advances to the next input.

If the feature information for the proper name is unknown at step 4404 , the default feature information is provided ( 4406 ). The default feature information is provided for a proper name whose feature information is unknown, as shown in the lower part of FIG. 36. There, for example, the proper name "Johnson" of the feature information is unknown and is provided with all types of feature information, ie "person, place, group and other". By providing the proper name, whose feature information is unknown, with all types of feature information, it is possible to leave a space so that the proper name can be broken down into one of the many features in the subsequent structural analysis.

If at step 4404 the feature information for the proper name is not known, the flowchart advances to step 4408 and the pointer is advanced to another input. It is then judged whether the input indicated by the pointer is at the end or not ( 4408 ), and if it is not at the end, the flow returns to step 4402 and the next input is checked to see if it is on Proper name is or not. If the input indicated by the pointer is at the input, the provision of the default information is ended. After the provision of the default feature information on the proper name is finished, the data recorded in the table 4036 a is released from the section 4036 to the section 4036 for parsing ( 4124 ), thereby performing the morpheme analysis therein Execution is complete.

The operation of the device described above will now be explained, for example, on an input sentence. The explanation is given in the event that a sentence "met at Walter Bahnhof Tokyo mr. Walter Johnson"("In Tokyo Station Mir. Walter met Johnson") is entered. First, input processing ( 4100 ) is performed to read the input sentence into the processing section 4014 . Then the dictionary reference unit is divided ( 4102 ) and the input sentence is divided into words by spaces. The reference dictionary 4020 is first retrieved for "on"("In") ( 4106 ). Since there is no "up" input located in the reference dictionary 4020 , the step first goes to processing for the proper name not registered in the dictionary. However, since the previous part is at the top of the string entered, it is considered a candidate for the end of the sentence that has been retrieved as "on or in" by adding I or A in a or i from the reference dictionary 4020 is implemented. Since there is no proper name ( 4110 ), the data retrieved from dictionary 4020 is recorded in table 4036 a ( 4112 ). Dictionary 4020 is then retrieved regarding "Tokyo" ( 4106 ). Since there is no entry in dictionary 4020 for "Tokyo" ( 4108 ) and the first character is a capital letter ( 4116 ), processing is performed for the proper name not registered in the dictionary ( 4120 ). Then, the flowchart in Fig. 34 is advanced. Since the previous part is "On or In", this is not a candidate for the end of the sentence ( 4300 ); "On" or "In" is not judged to be the end of the sentence ( 4312 ) "Tokyo" is recognized as a proper name whose feature information is unknown ( 4314 ) and processing is carried out for the proper name registered in the dictionary ( 4316 ) .

The flowchart advances to Fig. 33. Since "On or In" in the previous part is neither an unregistered ( 4200 ) nor a registered proper name ( 4204 ), it is recorded as a proper name which has inherent characteristic information, ie as a proper name, the characteristic information of which is unknown is ( 4216 ). In Fig. 32, the retrieval for dictionary 4020 for "station" is then performed ( 4108 ). Since there is an entry in dictionary 4020 for "station" ( 4108 ) and since it is a proper name ( 4110 ), processing is performed for the proper name registered in the dictionary ( 4114 ). The flow chart of FIG. 33 will now be transitioned to. Since "Tokyo" is an unregistered proper name in the previous part ( 4200 ), "Tokyo Station"("TokoyStation") is recorded as a whole as a proper name, which contains the feature information "location and group" in the form of the expression "station" (or "station") has ( 4202 ). Then the reference dictionary 4020 in Fig. 32 for "Mr." wrestled ( 4106 ) because "Mr." is an entry in dictionary 420 and a proper name ( 4110 ), processing is performed for the proper name registered in the dictionary ( 4114 ). Then the flow advances to Fig. 33. "Station" in the previous part is not an unregistered proper name ( 4200 ), but a registered proper name 4204 ) and the feature information (location, group) are not unknown ( 4206 ). Because "Mr." feature information is "person" and is unknown ( 4208 ), it is judged whether any common part in the feature information is "station" in the previous part and "Mr." exists ( 4212 ). Since "station" means "place, group", while "Mr.""Person" means and there is no common part between them, "Mr." registered alone as a proper name with the feature information "person". ( 4216 ). The flowchart in FIG. 32 is then returned to and the reference dictionary 4020 is retrieved for "Walter" ( 4106 ). Since there is an input for "Walter" in the dictionary ( 4020 ) ( 4108 ) and this is a proper name ( 84110 ), processing for the registered proper name is performed ( 4114 ). The flowchart of FIG. 33 is then returned to. Because "Mr." in the previous part is not an unregistered proper name ( 4200 ) but a registered proper name ( 4204 ), and since the feature information "person" is not unknown ( 4206 ) and furthermore the feature information for "Walter""person, place group" is also not unknown the common part for the feature information is judged ( 4212 ). Since there is a separate part ("person") that is available for the feature information, "Mr. Walter" is recorded together as a common noun with the feature information "person" ( 4214 ).

Then "met" is retrieved from dictionary 420 . And since there is an input ( 4108 ) and it is not a proper name ( 4110 ), the data recorded from dictionary 4020 are recorded in table 4036 a ( 4112 ). "Johnson" is also retrieved from dictionary 4020 ( 4106 ). Since there is no input for "Johnson" ( 4108 ) and the first character is an uppercase character ( 4116 ), processing is performed for the proper name not registered in the dictionary ( 4120 ). The flowchart of FIG. 34 is then transferred to. Since "met" in the previous part has no candidate for the end of the sentence ( 4300 ), it is judged whether "met" is not the end of the ( 4312 ), "Johnson" is considered a proper name with unknown feature information ( 4314 ) and processing is performed for the proper name registered in the dictionary ( 4316 ). The flowchart of FIG. 33 is then returned to. Since "met" does not exist as a non-registered ( 4200 ) or registered proper name ( 4204 ) in the previous part, "Johnson" is recorded alone as a proper name whose feature information is unknown.

After the previous processing step, default feature information is provided for the proper name, as shown in FIG. 35. The pointer is set to "In" as the upper end of the dictionary reference unit ( 4400 ). Since it is not a proper name ( 4402 ), the pointer is advanced ( 4408 ) and set to "Tokyo Station"("TokyoStation"). Since "Tokyo Station" is a proper name ( 4402 ) and the feature information is not unknown, since the entire part "Tokyo Station" was recognized as a place, group in the previous processing for the registered proper name ( 4404 ), the pointer is advanced ( 4408 ) and set to "Mr. Walter".

Since "Mr. Walter" is also a proper name ( 4402 ) and the feature information is not unknown ( 4404 ), the pointer is advanced ( 4408 ). Since "met" is not a proper name ( 4402 ) the pointer is advanced ( 4408 ). Since "Johnson" is a proper name ( 4402 ), the feature information of which is unknown, a default feature information is provided ( 4406 ) and "Johnson" is provided with feature information "person, place group, etc." provided as shown in Fig. 36.

As described above, in this embodiment, a German is divided (English) input sentence in retrieve character strings, which are then retrieved from the reference dictionary 4020, and if there is an input as a proper name in the dictionary 4020, a processing for the registered Proper names performed. When processing the registered proper name, the previous retrieval string is taken into account, and if it is a proper name, the feature information for the previous retrieve string and proper name as the object is checked. If it has no feature information, the other feature information is provided, while if there is any feature information for both, the common part is regarded as the feature information of these proper names. Consequently, it is possible to appropriately provide a proper name with no feature information with feature information and to restrict the provided feature information to more suitable feature information. This enables a more effective analysis in the subsequent structural analysis and when performing a corresponding translation.

Further, if the character string not registered in the dictionary 4020 , the first character is an uppercase character and it is determined that the previous character string is the end of the sentence because the uppercase character is converted to an uppercase character and dictionary 4020 is retrieved, it is possible to also retrieve the string at the top of the sentence in dictionary 4020 . Further, when a character string beginning with an uppercase character appears in a part other than the upper part of the sentence, it is judged as a proper name and the characteristic information for the proper name is provided with registered characteristic information by means of a proper name if it before and after exists. As a result, a proper name that is not registered in the dictionary 4020 can be broken down grammatically to some extent.

Furthermore, since a proper name does not have any feature information  is provided with all necessary feature information is provided and the unnecessary feature information removed in processing for the word is, it is possible to use a proper name whose characteristic information is not known or an unregistered proper name analyze.

Furthermore, since there is a number of feature information for a whole certain proper names are provided and suitable characteristic information depending on the feature information of the proper name before and after, it is possible to provide appropriate feature information for select the case that a proper name with a number of feature information in relation to others before and then broken down grammatically to thereby to enable effective division of the entered sentence.

A fifth embodiment of the invention will now be described, and FIG. 38 shows the entire structure of the fifth embodiment of a language analyzer according to the invention, which is applied to an automatic translation device for German (English) -Japanese. This embodiment has an input section 5010 , and German / English text 5020 to be translated into Japanese is entered through the section. The input section 5010 may include , for example, a keypad with character keys such as alphanumeric or function keys, an optical character recording device (OCR) to read the English / German text recorded on paper, and / or a file storage device for reading the German / English text, which is recorded on a storage medium, such as a magnetic disk.

The German / English text input from the input section 5010 is read into a pre-editing section 5014 , where pre-treatment for the translation is performed. In this case, sentence recognition and processing of unknown words are mainly performed. These functions are part of the morphological analysis. The pre-edited German / English data along with the information obtained from the pre-editing are transferred to a section 5016 for morphological analysis. Section 5016 divides the data into sentences while a word / dictionary 5018 is fetched, breaks down German / English morphemes, processes for unknown words, proper names, various compositions such as a time expression, and performs processing for the entire sentence, like an additional question (tag question) and a recognition of position. The rules for the morphological analysis are stored in a rules file 5036 .

After the morphological analysis, the German / English data are transmitted to a section I 5020 for syntactic analysis together with the dictionary information obtained in the morphological analysis. Section I 5020 is a functional section that grammatically breaks up the surface structure for the sentence by applying grammatical rules to the German / English data, and it finds out every structural possibility.

The German / English data which have been subjected to the syntactic analysis in section I 5020 are given together with the analyzed information to a section II 5022 for syntactic analysis, whereby a solution is obtained from the result of the syntactical analysis with regard to the surface structure the syntactic analysis I is selected using the structural description. A plausible parsing tree for the German / English sentence is therefore prepared and its structure is formed. The rules for a syntactic analysis are also stored in the rules file 5036 .

The German / English data are transmitted to the structure conversion section 5024 as the data of the so-called parsing tree after the syntactical analysis has been carried out. In section 5024 , a structure tree of the corresponding Japanese sentence is prepared from the structure tree, which is an intermediate structure of the German / English sentence, and is converted into a structure underlying the Japanese, from which the Japanese sentence can be easily translated. The data for the structure tree, which indicates the underlying Japanese structure that has been subjected to the structure conversion, is supplied to a translation section 5026 , in which a translation replacement is generated. This is a functional section that creates a Japanese sentence from the tree structure of the Japanese structure tree.

Japanese data as a result of the translation, that is, data for the translated sentence, is then given to a post-editing section 5030 . Section 5030 modifies the translated data by retrieving a dictionary 5018 using information that has been used in the translation to perfect a more natural Japanese sentence. The data for the Japanese set is output to an output section 5032 and as a translated Japanese set 5034 from the output section 5032 . The dispensing section 5032 may include, for example, a printer, a display, and a file storage device such as a magnetic disk. A flow of such a series of translations is controlled by a control section 5038 which controls the control of the entire facility. The word dictionary 5018 stores dictionary data for German / English and Japanese words, and various information is set therein, such as a connecting relationship, that is, an existing relationship, as well as meanings, a singular or plural form, a language part, etc., and all of this in addition to that Vocabulary. A file 5036 also stores control data for morphological and syntactic analysis.

The control section 5038 is connected to an operation display section 5040 which has operation keys to provide various displays from an operator to the device of the present invention, such as a translation display key, a cursor key, etc., and a display or a display which is visual one German / English text entered, a Japanese text as the result of the translation, intermediate data such as dictionary information, various displays for the operator, etc. It can also be implemented such that most of the operating display functions are provided in a keypad when it is arranged on the input section 5010 or in a display when it is arranged on the output section 5032 .

In Fig. 37, detailed structures of the processing of proper names in the portion 5016 are shown for a morphological analysis on. Section 5016 is shown for the part which is directly related to the understanding of the invention, although there are of course other functional sections relating to morphological analysis. The morphological analysis is performed by commanding dictionary retrieval from the top of the input string one by one in accordance with the retrieval key string, and processing for the dictionary information thereby obtained is performed by dictionary retrieving section 5104 in accordance with the position information of the proper name, which will be described later.

Section 5016 has an input processing section 5100 for receiving the data for the string input from section 5014 and performing the input processing. The section 5100 is provided with a buffer for the input string, which is supplied with the German / English string data in the form of code data, for example ASCII, and temporarily stores the string data.

The data for an input string that is temporarily stored in section 5100 is provided to section 5102 , which divides the data into dictionary reference units, such as words. Section 5102 is a functional section that distinguishes the dictionary reference units that represent the retrieval key strings to successively retrieve dictionary 5018 in section 5104 . The dictionary reference boundaries, which are used in the division for the dictionary reference units, are arranged at the respective position of the German / English character as a number, apostrophe, characters other than hyphen and period and as an apostrophe which follows a space. They are stored in a delimitation table 5108 and are referenced when the dictionary reference units are divided in section 5102 .

The reference dictionary 5018 particularly stores the information for retrieving the partitioning unit. For example, as shown in Fig. 38 for the example of the input information, grammar information such as a sentence for each of the dictionary reference units is included, for example, an input for a word. The language part information contains a hint for the noun if it is a general noun or a proper name. For the proper name, a distinctive note which shows the way to limit the position in the sentence, ie position information for the proper name, is stored. This will be described in detail later. Other items of information include, for example, a countability or non-countability of a noun, a distinction such as an intransitive or transitive verb, its translation, etc.

There are four types of position information for the proper name, d. H. in the present embodiment, "0" to 3". The pattern "0" shows a proper name without position restriction, for example "Stadt / City" or one personal name "Walter". The pattern "1" indicates that it's a proper name, for example "Mr." is the one at the top End of a single proper name is arranged, or one Sequence of a number of proper names, i.e. H. a proper name, which is arranged in a single group of proper names. The pattern "2" indicates that it is a proper name, for example "Bahnhof / Station" or "Bucht / Bay", which end of a single proper name are arranged, or that it is a Proper name is that in a word as a group of proper names is arranged, or which other than the pattern "3", which will be described below. The pattern "3" shows assumes that it is a proper name, for example "River" in "the Sumida River" which is the same as pattern "2" however, from a certain article "the / the" at the beginning of a Proper name of the in a proper name as a group of Proper name is arranged.

The dictionary retrieval section 5104 is a functional section which outputs information by retrieving the word dictionary 5013 based on the retrieval key string entered in the section 5102 and transmits the same to the dictionary information preservation table 5124 , to the position information processing section 5110 and to section 5111 which judges the previous sentence end. The processing based on the patterns "0" to "3" corresponds to the position information for the proper names retrieved from the dictionary 5018 are performed by proper name processing sections 5114, 5116 and 5118 . Processes for proper names are performed using the pattern "1" in the section 5114 , the pattern "2" and "3" in the section 5116 and the pattern "0" in the sections 5118 and 5114, respectively.

In this embodiment, proper names are arranged collectively, for example, using a word as a key that is part of a group of proper names grouped in a single proper name, and positionally constrained when arranged as a proper name. Even if a number of proper names are continuously present, they can be appropriately arranged together with the context without such an erroneous composition by simply always being arranged as a single group of proper names. Processing for this purpose is performed in sections 5114, 5116 and 5118 .

Proper names with a certain scope are registered in the reference dictionary 5018 . Such proper names registered in the dictionary are subjected to a grammatical breakdown or morphological analysis in section 5110 and sections 5114, 5116 and 5118 . They form a processing section for proper names registered in the dictionary. The proper names that are not registered in the dictionary 5018 are parsed or morphologically analyzed in the section 5112 which judges the preceding sentence end and in the section 5118 for processing a proper name corresponding to the pattern "0". These form the processing section for proper names not registered in the dictionary.

Processing a proper name is carried out in the following two steps. First a proper name is recognized in the entered string. In the case of a word registered in dictionary 5018 , this is done by displaying the proper name in its morpheme actuation information. In the case of a word that is not registered in the 5018 dictionary, this is done so that a character at the top is a capitalized German / English character, for example "John" or "US" etc.

Then a group of proper names is arranged collectively, to make the whole part a single proper name. If recognized as a proper name from the dictionary information and when the next dictionary reference unit is also a proper name, the whole part becomes collective composed of a proper name. For example, "M. Weber "as a whole was analyzed as a proper name. The result The analysis forms a candidate for the ideomatic Expression including proper names in the local group grammatical breakdown or analysis.

Then the necessary local analysis is carried out. In in this case, a sequence of so-called parsing Units identified by the morpheme actuation information operated for each of the parsing units, collectively arranged in a parsing unit, what's on a local Parsing rule is based. For example, "Mr. Brown" arranged in "brown shi". Furthermore, words become one Represent part of a district name also arranged collectively. For example, "Lake Biwa" is put together in "Biwako". In the same way, words that are part of a group name, also arranged collectively. For example, "Yale University" is called "Yale Daigaku" analyzed.

In the case of the proper name "Mr...." and "Lake..." there is always an end just before that in terms of context. Consequently, if "Tom Brown" is collectively combined into a single proper name, an error is caused in the following analysis. For example, a proper name "Universität / University" always comes to an end immediately afterwards. For example, in an English sentence "At Yale University Tom is ..." it is recognized that there is an interruption between "University" and "Tom". In this embodiment, information for the position at which corresponding proper names in the sequence of proper names are restricted in position is stored in the dictionary 5018 as the above-described position information, that is, as a pattern "0" to "3". The collective arrangement with this position information is carried out in the processing sections 5110, 5112, 5114, 5116 and 5118 . The dictionary information for the input string after the completion of these operations is stored in the retrieved dictionary information buffer, that is, the dictionary information preservation table 5124 .

The result of the morphological analysis is transferred from table 5124 to section I 5020 for morphological analysis. The processing by the proper name position transfer is carried out using the sequence shown in FIG. 40. Input processing is performed by including the data for the input string in the input processing section 5100 ( 5200 ). Then section 5102 divides the input string into dictionary reference units to retrieve dictionary 5018 ( 5201 ). Section 5104 retrieves dictionary 5018 accordingly ( 5203 ), and if there is a dictionary entry ( 5204 ), it checks its speech portion ( 5205 ). If the speech part is not a proper name, processing for the proper name is not performed in this embodiment because the dictionary information is stored in table 5124 ( 5206 ). If it is a proper name, processing for dictionary name 5207 registered in the dictionary is performed in section 5110 and sections 5114, 516 and 5118 . If this processing is carried out at the end position of the sentence which is indicated by the data of the input character string ( 5202 ), the result of the morphological analysis is delivered to the section I 5020 provided for this purpose ( 5210 ).

If there is no input at step 5204 as a result of the dictionary reference and if the element starts from an uppercase character ( 5212 ), this is recognized as a proper name that is not registered in the dictionary and processing for one not in the dictionary Registered proper names 5213 are performed in the previous part judgment section 5112 and the proper name processing section 5118 . If the initial character is not an uppercase character because this is a word not registered in dictionary 5018 , it is stored as an unregistered word in table 5124 ( 5214 ). Processing is then performed at the end position ( 5202 ).

The processing for the proper name 5207 registered in the dictionary is performed in the processing sections 5110, 5114, 5116 and 5118 in the flowchart shown in FIG. 41. First, the obtained position information contained in the dictionary information is related ( 52209 ). Proper name processing 5221 for pattern "0" is performed when it indicates pattern "0"; proper name processing 5220 is performed for pattern "1" if it indicates pattern "1", and proper name processing 5223 is performed for pattern 2,3 if it indicates pattern "2" or "3" . Processing 5221 for pattern "0" is performed in section 5114 . The processing is applied to a proper name that has no position restriction. First, if a part that precedes the dictionary reference unit in question is an unregistered proper name ( 5230 ), the whole part is collectively arranged in a proper name where the position information is "1" and is stored in table 5124 ( 5233 ) . If the previous part is a proper name with position information "1" 852319 , the processing is carried out in the same way.

If the previous part is a proper name with position information "0" 85231 ), the whole part is collectively put together to form a proper name with position information "0" and stored in table 5124 ( 5235 ). Further, if the preceding part is not a proper name with position information of "0", the whole part alone is stored as a proper name with position information "0" in table 5124 ( 5134 ).

Proper name editing 5222 for pattern 1 is performed as described below. Editing is applied to a proper name, such as "Mr.", which is placed at the beginning of a single proper name or at the beginning of a proper name arranged in a group as a sequence of a number of proper names. First, if the portion preceding the relevant reference unit is an unregistered proper name ( 4240 ), the word is converted to unregistered ( 5241 ). If it is not an unregistered word, it is stored alone as a proper name with position information pos "1" in table 5124 ( 5242 ).

The proper name processing 5523 for the patterns 2, 3 will now be described with reference to FIG. 44. The processing is applied, for example, to a proper name such as "station" or "river" arranged at the end of a single proper name, or to a proper name that is in a group as a series of numbers is arranged by proper names. First, if the part preceding the relevant reference unit is an unregistered proper name ( 5250 ), this part is collectively put together with the previous word as a proper name with its own position information pos-self as the proper name position information pos and in that Table 5124 saved ( 5225 ). Further, the processing is carried out in the same manner when the previous part is a proper name with the position information "1". ( 5251 ).

If the previous part is not a proper name with position information "0", ( 5252 ) it is arranged alone in a proper name with a single position information pos-self as the proper name position information and stored in the table 5124 ( 5257 ).

If at step 5252 the previous part is a proper name of the position information "0", its own position information pos-self is checked ( 5232 ) and processing 5255 is performed if it is the pattern "2". If the own position information pos-self is the pattern "3", it is further checked whether the element in front of a reference unit is "the"("the") or not. If it is not the particular article "that", processing 5255 is performed. If it is "the", the group of "the" with its own element is collectively assembled as a proper name with the position information "3" and stored in the table 5124 ( 5256 ).

For a word that begins with an uppercase letter and is recognized as an unregistered word for which there is no input from dictionary 1518 as a result of retrieval 5203 , processing 5213 is performed in steps 5204 and 5212 in the flow diagram and processing 5213 is performed in the previous sentence end judging section 5112 . First, if the part preceding the relevant reference unit is not a candidate for the end of sentence, processing 5221 for pattern 0, as described above, is performed in processing section 5118 .

The previous part can be a candidate for the end of the sentence in the following four cases. The first is a case at which a separate point "." is available. The next is a case in which the previous input the point is in the last position, and the position information  for the proper name is not 21 ". This case includes, for example, an abbreviation such as "U.S.A." a. There is also the case of a colon ":", a semicolon ";", a sequence of a period and an apostrophe Write ". '" and a sequence of one point and one Quotation marks "." ". The latter case is the case that it is at the top of the buffer for an entered String.

If the previous part is one of the four cases described above, the previous sentence end candidate is recognized as the sentence end ( 5261 ) and the dictionary is retrieved after the uppercase letter of the word is converted to a lowercase letter ( 5262 ). If a dictionary input is obtained as a result of the retrieval ( 5263 ), it is recorded in table 5124 ( 5264 ). If not, it is recorded as an unregistered proper name in table 5264 , leaving the upper character unchanged as the uppercase letter ( 5265 ).

Another example will now be explained. For example, if a dictionary entered a string "Along the Sumida River, Paul and mr. Gold Smith..."("Along the Sumida River Paul and Mr. Gold Smith went..."), The dictionary input information is first written to table 5124 , as shown in Fig. 64A. For example, for "des" and "the "the starting position in the sentence" 7 "and the ending position is" 9 ", and the language part is a specific article. No entry can be made for the word" Along "at the top of the entered string in the dictionary- Fetch 5203 is obtained and it is determined to be unregistered, however, since the previous part may meet the candidate's end-of-sentence requirement because it exists at the top of the input buffer ( 5260 ), the capital letter "E""A" is initially converted to a lower case letter and dictionary retrieval 5262 is performed as "along""along".

Then the pointer is incremented to proceed to processing for "Sumida". This word is not registered in dictionary 5018 in the present embodiment. Since the previous part is not a candidate for the end of the sentence, the flow is transferred to proper name processing 5221 for pattern 0. As shown in Fig. 46A, a proper name for the voice part and "0" for the proper name position information are retrieved.

The next reference unit "river""river" is a proper name with the position information "3". The previous part is a proper name with position information "0", and the other previous part is "des"("the"). In view of this, the expression "the Sumida River" is collectively compiled as a single proper name by steps 5250 to 5254 and stored as the position information "3" in table 5124 ( Fig. 46B). Then the next reference unit "P" is an unregistered proper name with the positioning information "0", on which processing 5213 is carried out. Although the previous word is a proper name because the position information for this is "3", this does not collect anything collectively, but rather the dictionary information is stored as is in table 5124 ( FIG. 46C). Ordinary processing is used for the subsequent conjunction "and".

The next word "Mr." is a proper name with position information "1" which is stored as it is in the table 5124 ( Fig. 46D). Even if the previous part is a proper name "Paul", punctuation between words can be directly from "Mr." exists as well as it is stored in table 5124 .

Furthermore, the word "gold" is a proper name that is not registered in the dictionary and to which processing 5213 is applied. Since the previous word "Mr." having position information "1", both are collectively combined, and the whole part is formed in a proper name with the position information "1" ( Fig. 46E). Then the processing on the next word "Smith" is similar to " Fig. 46F." The subsequent word "went" is a past tense of a verb and a common grammatical breakdown or analysis is performed below.

As above with reference to the inventive Embodiment is described, proper names thereby arranged that a word is used as a key that forms part of a group of proper names, which collectively combined into a single proper name, and that goes through a position restriction when they collectively in a proper name. Self if there is a constant number of proper names, it is in this way possible a correct collective compilation along with the context without a flawed perform collective compilation by simply copied in a group of proper names. In the aforementioned Example the word "of the Sumida river" ("the Sumida river") analyzed as a group of proper names, which are separated from the following word "Paul" are. Furthermore, "Mr. Gold Smith" is featured as a group Proper names analyzed.

A sixth embodiment of the speech analyzer according to the invention, which is used in an automatic translation device English / German-Japanese, will now be described with reference to FIG. 47. This embodiment has an input processing section 6014 , and data is input to section 6014 from an input unit 6012 . The input unit 6012 has, for example, a keypad with character keys, such as alphanumeric and function keys, an optical character reading device for reading the English / German text recorded on paper and a reading device for a magnetic disk.

The input processing section 6014 has a buffer 6014 a for an input character string and stores the English / German sentence entered by the device 6012 in the buffer 6014 a . The section 6014 reads out the set stored in the buffer 6014 a and delivers it to a unit division section 6016 . Section 6016 is a functional section that divides the dictionary reference unit from the input set from section 6014 by retrieving a delimitation table 6018 . Table 6018 contains delimitations, such as spaces, commas, etc.

Section 6016 reads the boundaries from table 6018 and divides the sentence entered from section 6014 into strings as the units for querying a reference dictionary 6020 by dividing the sentence into parts where the boundaries exist. The split strings are input to a dictionary retrieving section 6022 .

Section 6022 retrieves the reference dictionary 6020 for the input sentence output from section 6016 , which is divided into reference units. For example, the reference dictionary 6020 contains, as shown in Fig. 48, inputs for the character strings, their language part, feature information, etc. of the English / German sentence. The reference dictionary 6020 contains, in addition to the proper names shown in the figure , the strings for another language part such as verbs, adjectives, etc. The proper name as a language part in this figure means that they are used in connection with the registered proper name distribution, what will be described later, but does not express a common grammatical proper name. Furthermore, the feature information shows what the relevant proper name expresses and is not limited to just one.

Section 6022 retrieves the reference dictionary 6020 for the string divided into the reference unit, and if the string is a proper name, it passes it to a section 6024 to perform the proper name processing, which will be described later. Furthermore, if it is not a proper name, it is given to a processing section 6036 and is stored in the table 6036 a in the processing section 6036 . Section 6024 has a section 6026 which processes the preceding sentence end, a section 6030 which processes the previous proper name.

In section 6026 , it is judged whether or not a string preceding the input string to be retrieved from section 6022 is at the end of the sentence, and when the previous string is at the end of the sentence, it is given to section 6022 after the uppercase letter at the beginning of the string to be processed is converted to a lowercase letter, thereby causing section 6022 to be classified as an unregistered proper name in the string that was not retrieved by the second retrieval submitted the processing section 6036 and kept in its table 6036 a . If the character string preceding the input character string is not at the end of the sentence, it is given to the processing section 6036 as a proper name, the feature information of which is unknown, and is registered in the table 6036 a , as will be described later.

Section 6028 decomposes the feature information for the previous character string output by section 6026 and delivers the result to section 6030 . Section 6030 processing the proper name per se checks the feature information for the proper name to be broken down or analyzed, and as will be described later, if the feature information is not registered as either the proper name or the previous proper name, it analyzes the proper name and the previous proper name with the help of the registered feature information of the other proper name and stores this result in the table 6036 a in the processing section 6036 .

The processing section 6036 has the dictionary information preservation table 6036 a, stores the data output from the section 6028 or 6032 in the table 6036 a and then reads out stored data in this way and delivers it to a section 6038 which carries out a parsing , which carries out the structural analysis for the entered sentence , reads it from table 6036 a and subjects it to a morphological analysis.

The operation of the device will be explained with reference to the flow chart shown in FIG. 49. First, a German / English input sentence is read from input device 6012 into input processing step 6014 ( 6100 ). The read section 6014 in the input sentence is stored in the buffer 6014 a, and a stored in the buffer 6014 is read out to the input sentence units- Aufteilabschnitt 6016th

When the input sentence is read in, section 6016 reads delimitations from the corresponding table 6018 to perform a division for the dictionary reference units ( 6102 ). That is, the strings representing the input sentence inputted are sequentially divided into fetch key strings as the units when fetching the reference dictionary 6020 from the beginning thereof by dividing them at the places where the delimitations such as spaces and colons, available. It is judged whether the divided reference units, that is, the retrieval key strings, have ended or not ( 6104 ), and if there is still a (unfinished) retrieval string, it is delivered to the dictionary retrieval section 6022 .

When the fetch string is provided to section 6022 , this ( 6022 ) fetches the reference dictionary 6020 for the fetch string ( 6106 ). It is then judged whether or not the retrieval string is present in the input of the reference dictionary 6020 , as shown in Fig. 48 ( 6108 ), and if there is an input, part of speech information stored in the reference dictionary 6020 is stored is read out, and it is judged whether the retrieval string is a proper name or not ( 6110 ).

If the retrieval string is not a proper name, the section 6022 passes the data read from the reference dictionary 6020 to the processing section 6036 and stores it in the table 6036 a ( 6112 ). When the data is stored in the table 6036 a , an input indicating that the data is stored and the data for the retrieval string which have been stored immediately before are input to the section 6016 by the processing section 6036 . Thus, step 6012 is returned to, and the division for dictionary reference units is performed in section 99999 00070 552 001000280000000200012000285919988800040 0002003733674 00004 99880nitt 6016 .

If, in step 6110, the retrieval string is a proper name, section 6022 returns the proper name read from reference dictionary 6020 (hereinafter referred to simply as a proper name) along with the data of the previous one entered from table 6036 a Retrieve string using section 6016 to dictionary query section 6022 and section 6024 where processing for the proper name registered in the dictionary is performed ( 6114 ).

Now, with reference to the flowchart shown in Fig. 50, processing for the proper name registered in the dictionary is performed. The data which have been discharged from the section 6022 to the section 6024 are delivered to the proper names previous processing portion 6028 by means of the previous block end processing section 6026th Section 6026 has no function when processing the proper name registered in the dictionary.

In section 6028 , it is then judged whether or not the retrieval string preceding the proper name is a proper name not registered in the reference dictionary 6020 , that is, whether it is a proper name that is subject to processing for the proper name not registered in the dictionary ( 6200 ), or not, as will be described later. If it is an unregistered proper name, the entire part of the proper name and the previous unregistered proper name are judged as a proper name having the characteristic information of a proper name ( 6202 ); the data is then passed to the processing section 6036 and stored in its table 6036 a ( 6214 ).

In the processing section 6028 , if the retrieval string preceding the proper name is judged to be an unregistered proper name, it is judged at step 6200 whether or not the retrieve string preceding the proper name is a proper name registered in the reference dictionary 6020 ( 62049. If the retrieval string preceding the proper name is a registered proper name, it is judged whether or not the feature information of the previous proper name is unknown, that is, whether it is not registered in the reference dictionary 6020 or not ( 6206 ).

If the feature name of the previous proper name is unknown, the flowchart advances to step 6202 where the entire portion of the proper name and previous proper name is considered a single proper name that has the feature information of the proper name ( 6202 ); the processing section 6028 then outputs data from the processing section 6036 , where it is stored in its table 6036 a ( 6214 ).

In the processing section 6028 , if the feature information of the previous proper name is not judged to be unknown, that is, if it is judged that it is registered in the reference dictionary 6020 , the data is released from the processing section 6028 to the section 6030 . It is then judged in section 6030 whether or not the feature information of the proper name is known ( 6208 ). In the event that the characteristic information for the proper name is unknown, the section 6030 judges the whole part of the proper name and the previous proper name as a proper name which has the characteristic information of the previous proper name ( 6210 ) and delivers the data to the section 6036 . In its table 6036 a they are recorded ( 6214 ).

If it is determined in the section 6030 that the feature information of the proper name is not unknown, that is, that it is registered in the reference dictionary 6020 , the processing section 6030 judges the proper name as a proper name with feature information which is extracted from the reference dictionary 6020 independently previous proper name has been retrieved ( 6212 ) and transfers the data to the processing section 6036 , in whose table 6036 a the data are then recorded ( 6214 ). . Turning now back to Figure 49 no retrieval string is present at the input of the reference dictionary 6020 in step 6108, it is judged whether the first character in the string is an uppercase letter or not (6116); if it is not an uppercase letter, section 6022 judges the fetch string as an unregistered word, passes it to section 6036 and stores it in table 6036 a ( 6118 ).

If the first character is an uppercase letter, the data for the fetch string is given together with the data for the previous fetch string from the section 6022 to the section 6024 in which the processing for an unregistered proper name is performed ( 6120 ).

The processing for a proper name not registered in the dictionary will now be described with reference to FIG. 51. The data for the fetch string, together with the data for the previous input recorded in the table, is delivered to the previous sentence end processing section 6026 when judging whether the end of the previous input recorded in the table is a candidate for the end of the sentence or not ( 6300 ). This judgment on the candidate for the end of the sentence is made to judge whether or not the end of the previous entry recorded in the table is a candidate for the end of the sentence, e.g. a separate period (.), Etc.

If the end of the previous entry recorded in the table is a candidate for the end of the sentence, data is delivered from section 6026 to section 6028 processing the previous proper name, and section 6028 then judges the previous entry recorded in the table as that End of sentence ( 6302 ) and outputs to dictionary retrieval section 6022 after the uppercase letter at the beginning of the retrieval string has been changed to a lowercase letter.

Section 6022 retrieves the reference dictionary 6020 for the fetch string that has been refurbished for the lower case letter ( 6304 ) and judges whether it is an input to the reference dictionary 6020 ( 3603 ). If there is an input, section 6022 passes the data retrieved from reference dictionary 6020 to processing section 6036 and stores it in table 6036 a ( 6308 ). If there is no input, section 6022 returns to the first character in the fetch string on upper case, passes it to processing section 6036 as an unregistered proper name, and stores it in table 6036 a ( 6310 ). If at step 6300, the processing section 6026 judges that the end of the previous recorded in the table input is not a candidate for the end of the block, the data is output from the section 6026 to the processing section 6028 and the portion 6028 then judges that the in previous entry recorded in the table is not the end of the sentence ( 6312 ). The data is then passed from section 6028 to section 6030 , and this section 6030 then judges the retrieval string as a proper name whose feature information is unknown ( 6314 ).

Section 6030 then returns to the data in section 6028 and processing for the proper name registered in the dictionary is performed in section 6028 ( 6316 ). The processing for the proper name registered in the dictionary is the same as that shown in Fig. 50.

If now in Fig. 49, the dictionary reference units are completed in step 6104, the given in the table 6036 a recorded data from the section 6036 to the section 6038 (6122), whereby the morphological analysis of this embodiment is terminated in accordance with. The operation of this embodiment, which has been described above, will now be explained using an input sentence.

In the explanation, reference is made to FIG. 52, where, for example, an input sentence "In Tokyo Mr. Walter station..."("In Tokyo Station Mr. Walter ...") is entered. First, input processing 1100 is performed by reading the input sentence into the processing section 6014 . Then the dictionary division unit is made ( 6102 ) by dividing the input sentence into corresponding words by spaces. First, the reference dictionary 6020 for "Im" is retrieved ( 6106 ). There is no input for "Im" in the reference dictionary 6020 . If the step of processing for a proper name not registered in the dictionary has advanced because the previous part is recognized as the beginning of the sentence (the top of the file), "Im" is converted to "." has an input in the reference dictionary 6020 and is not a proper name ( 6110 ), the data retrieved from the reference dictionary 6020 are recorded in table 6036 a ( 6112 ).

Then the reference dictionary 6020 related to "Tokyo" is retrieved ( 6106 ). Since there is no input for "Tokyo" in the reference dictionary 6020 ( 6108 ) and the first character is an uppercase letter ( 6116 ), processing for a proper name not registered in the dictionary is performed ( 6120 ). Then advance to Fig. 51. Since the previous part is "Im or In" and is not a candidate for the end of the sentence ( 6300 ), "Im or In" is recognized that this is not the end of the sentence ( 6312 ); "Tokyo" is recognized as a proper name whose feature information is unknown ( 6314 ), and processing for a proper name registered in the dictionary is performed ( 6316 ). Then advance to Fig. 50. Since the preceding "in or in" is neither a registered proper name nor an unregistered proper name ( 6204 ), "Tokyo" is recorded alone as a proper name with its own information feature, ie as a proper name whose feature information is unknown ( 6216 ), then at the next step to Fig. 49 decreased, and the reference dictionary 6020 is retrieved for "station or station" (6106). Since there is an input for "station" in the reference dictionary 6020 ( 6108 ) and this is a proper name ( 6110 ), processing for a proper name registered in the dictionary is performed ( 6114 ). The next step advances to Fig. 50. Since the previous "Tokyo" is an unregistered proper name ( 6200 ), the entire part for "Tokyo Station or Tokyo Station" is recorded as a proper name that has the feature information "location, in the form of" station or station "( 6206 ).

Then the reference dictionary 6020 in Fig. 49 for "Mr." retrieved ( 6016 ). Since there is an entrance for "Mr." in the reference dictionary 6020 and there is a proper name ( 6110 ), processing is performed for a proper name registered in the dictionary ( 6114 ). Then, the next step is to advance to Fig. 50. The previous expression "station" is an unregistered proper name ( 6200 ), but a registered proper name ( 6200 ) and the feature information "location" is not unknown ( 6206 ). Because "Mr." the feature information "is not unknown" ( 6208 ), "Mr." registered alone as a proper name with the feature information "person" ( 6212 ).

The reference dictionary 6020 for "Walter" is then called up again in FIG. 49 ( 6016 ). Since "Walter" has an entry in the reference dictionary 6020 ( 6108 ) and it is a proper name ( 6110 ), processing is performed for a proper name registered in the dictionary ( 6114 ). Then move on to step 50 in the next step. Since the previous "Mr." an unregistered proper name is ( 6200 ) but a registered proper name ( 6204 ) with the feature information for "person" is not unknown ( 6202 ), while the feature information for "Walter" is unknown ( 6208 ), "Mr. Walter" is composed and recorded as a proper name with the feature information "person" ( 6210 ).

As described above, in this embodiment, the German / English input sentence is divided into fetch strings for which the reference dictionary 6020 is fetched first. If there is a proper name input in the reference dictionary 6020 , registered proper name processing is performed taking into account the previous input recorded in the table.

If the previous input recorded in the table is a proper name, the feature information for the previous input recorded in the table and checked for the current proper name to be processed. If any of these are missing from the feature information, the other's feature information is provided by them, while if both have feature information, they are individually used as proper names with associated characteristic information recognized.

As a result, it is possible to use a proper name that does not contain any feature information appropriately with feature information to provide, and the intended feature information limit appropriately. this makes possible a more effective analysis in the subsequent structural analysis and a corresponding translation.

Further, for the character string that is not registered in the reference dictionary 6020 when the first character is an uppercase letter and the previous character string is judged to be the end of the sentence, the reference dictionary 6020 is retrieved after the uppercase letter has been changed to a lowercase letter, and thus it is possible to get the 6020 reference dictionary for the string at the beginning of the sentence. Further, when a character string beginning with an uppercase letter appears in a place other than at the beginning of the sentence, it is judged as a proper name, and the characteristic information of the proper name is provided with a registered characteristic information by a proper name which is present before or after is. As a result, a proper name that is not registered in the reference dictionary 6020 can be parsed or analyzed to some extent.

A seventh embodiment of the invention will now be described. Here, 54 of the overall structure of the seventh embodiment is shown in Fig. A speech analyzer illustrated in accordance with the invention, which is used in a Japanese translation automtischen means for English /. This embodiment has an input notch 7010 through which an English text 7012 to be translated into Japanese is input. The input section 7010 may include , for example, a keypad with character keys such as alphanumeric or function keys, an optical character reader (OCR) that reads the English text recorded on paper, and / or a file storage device for reading an English text stored on a storage medium such as a magnetic disk is recorded.

The input from the input section 7010, English text is read in a Vorredigierabschnitt 7010, in which a translation pretreatment is carried out. In this case, sentence recognition and treatment of unknown words are mainly carried out.

This then acts as part of a morphological analysis. The pre-edited English data, along with the information obtained from the pre-edited, is transmitted in a section 7010 for morphological analysis. Section 7010 breaks down the morphemes of the English sentence by dividing them by retrieving a word dictionary 7018 , performs various types of arrangements or compilations through unknown word processing, a proper name expression, time, number, etc., and performs processing for the whole sentence, such as an additional question (tag question) and an apposition recognition. The morphological analysis rules are contained in a rules file 7036 .

The English data which have been subjected to the morphological analysis, together with the dictionary information which has been obtained by the morphological analysis, are transmitted to a section I 7020 provided for this purpose. Section 7020 is a functional section that grammatically breaks down the surface structure structure for the sentence using grammar rules on English data and finds all structural possibilities. The English data subjected to analysis in section I 7020 is then supplied together with the analysis information to section II 7020 serving for syntactic analysis, in which a solution to the result of the previous analysis corresponding to the surface layer structure in section I is applied a structural description is selected. As a result, a plausible parsing or parsing tree of the English sentence is prepared, and the structure is then made. The analysis rules are also stored in the rules file 7036 .

The English data is transferred to a structure conversion section 7024 as the data for the parsing or analysis tree after performing the syntactic or morphological analysis. This ( 7024 ) prepares a corresponding Japanese structure tree from the structure tree , which is an English intermediate structure, and converts it into a structure underlying the Japanese, from which a Japanese sentence can be easily translated.

The structure tree data showing the converted Japanese structure is given to a translation section 7062 , where a translated sentence is generated. This is a functional section for creating a Japanese sentence from the Japanese structure tree. The Japanese data, which is formed as a translated sentence, that is, the translated sentence data, is supplied to a post-editing section 7030 . Section 7030 modifies the translation sentence data related to dictionary 7018 using information that has been used in the translation to perfect a more natural Japanese sentence. The data for the Japanese sentence is transferred to an output section 7032 and then output from there as a translated Japanese sentence 7034 . The dispensing section 7032 may include, for example, a printer, a display, and / or a storage file device such as a magnetic disk. The flow of a series of translation processes is controlled by a control section 7038 which controls the control for the entire facility.

In this embodiment, the dictionary data for English and Japanese words are stored in the word dictionary 7018 . A vocabulary is defined, as well as a connecting relationship, ie a relationship which exists at the same time or various information, such as meanings of a singular or plural form, a language part, etc. Furthermore, rule data for a morphological and syntactic analysis are stored in the rule file 7036 . The control section 7038 is connected to the operation display section 7040 , which in turn has operation keys to issue various commands from an operator to the device, such as a translation command key and a cursor key, as well as a display or a display that visually shows the input Japanese Sentence text, the Japanese sentence as a result of a translation, intermediate data such as dictionary information displays various information to the operator.

Here, most of the operation display functions may be included in a keypad when it is disposed on the input section 7010 or in a display when it is disposed on the output section 7032 .

In FIG. 53, detailed embodiments for a morphological analysis by leading section 7016 are shown in relation to the processing of numbers. These parts, which contribute directly to the understanding of the invention, are shown, although section 7016 for morphological analysis naturally also has other functional sections for morphological analysis. Section 1017 has an input processing section 7100 for receiving and processing the input string data input from a pre-editing section 7014 . The section 7100 is provided with a buffer in which data for the English character string in the form of code data such as ASCII has been entered, and temporarily stores the data for the character string.

The input string data temporarily stored in the processing section 7100 is output to a section 7102 which divides the data into dictionary reference units such as words. Section 7102 is a functional section that distinguishes a dictionary reference unit that sequentially represents the string when dictionary 7016 is retrieved in section 7106 . Dictionary reference boundaries that have been used in the division processing for the dictionary reference unit are arranged in place of an English character, a numeric character, an apostrophe, characters that are not a hyphen or period, and an apostrophe that follows a space . They are stored in a delimitation table 7104 and are referred to in section 7108 when the dictionary reference unit is split.

The dictionary 7018 contains information, in particular for retrieving the split unit. The dictionary 7018 also stores morpheme-processing information such as the name of a month, a day of the week, a cardinal number which is only a numerical value, an ordinal number, a unit for expressing grams and the like. Time, the, of, comma (,). (.) etc.

The dictionary retrieving section 7106 is a functional section that retrieves the dictionary 7018 to take out the dictionary information based on the character string input from the section 7102 and transmits it to a morpheme processing information creating section 7108 . Section 7108 contains morphemic processing information (see Fig. 56) indicating that a string with a morphemic feature has a time meaning such as hour, year, month, etc., with further specific information provided in the recognized string to obtain a cardinal number or temporal meaning in the dictionary retrieving section 7016 . For example, such information is created that a numerical figure, ie a number, means "yes". The character string provided with information in section 7108 is also used in the case of a necessary local grammatical decomposition or analysis.

In this case, a unit group of the dictionary Reference unit, such as a word that is represented by the Morpheme actuation information has been actuated with Help from the local grammatical dissection or Parsing rules put together collectively. For example "Month Name", "Numeric Expression" are in "Month Name + numeric expression "compound, i.e." October "and "18" are grouped to "oct". 18 ". Also a collective compilation made, such as "November the 2nd" for "month name + the + numeric Expression "," 22 March "for" numeric expression + month name ", "the 23rd May" for "the + numeric expression + Month name "," the 11th of June "for" the + numeric expression + of + month name "," ′86, Jan. 27. Mon. "for" year +, plus month and day +, + day of the week ", Suday Jan 26, 1986 "for weekday +, + month and day +, + year", "11:30 am." for "number: number + a. m. (or p. m.)" or "Month name + year" "Month name + of + year" etc.

The processing for the local analysis is performed in an initial value setting section 7110 , an adjustment fetch section 7112 , a unit division section 7114 , a morpheme processing information generating section 7111 , query sections 7116 and 7120 and processing sections 7122 and 7124, and an adaptation table 7128 includes a morpheme processing display table which is a discrimination reference table to discriminate that a unit sequence showing a number and time factors as shown in Fig. 57 is a composite unit of time factors according to certain rules. The section 7110 sets the initial value of a counter n, which counts the number of dictionary reference units, to match a series of dictionary reference units in the matching query section 7112 when retrieved as a unit group described above.

Section 7112 calls the adjustment table 7128 for each of the dictionary reference units to perform an adjustment. The unit division section 7114 distinguishes the dictionary reference units which are assumed to be "p" , which has been completed with the dictionary query in section 7106 from the dictionary reference units which represent strings after the dictionary reference units with the help of the dictionary query have been completed by counter n .

The query section 7116 is a functional section which has a similar function to the dictionary retrieving section 6106 which retrieves the dictionary 7018 to take out the dictionary information based on the character string which has been distinguished in the section 7114 and the same on the morphemes Processing section 7118 . Section 7118 has the same function as section 7108 , with further specific information being provided for each piece of information which has been recognized as an ordinal number or a time factor in the query section 7116 .

The query section 7120 and the processing section 7122 and 7124 collectively assemble a series of dictionary reference units up to " p + n", which has been obtained from the adaptation retrieval section 7112 by the processing in the section 7118 , into a dictionary reference unit. Then the result is stored in table 7126 , which is a buffer for storing the dictionary information which has been completed by the retrieval. The result of the morphological analysis is transferred from table 7126 to section I 7020 for syntactic analysis.

The collective assembling by means of the morpheme processing information according to the invention will now be explained with reference to the flowcharts shown in Figs. 55A and 55B. For example, the following string should be entered in the input processing section 7100 ( 7300 ).
Entered string :. . "26 Jan., '80 he ..."
Section 7102 divides the input string by the dictionary reference unit for retrieving dictionary 7018 ( 7302 ) "26" in the input string is separated by separating the dictionary reference unit as a unit. It is now judged whether the separated part of the reference unit for the input character string has ended or not; if it is finished, the operation is ended ( 7304 ), while if it is not, the flowchart advances to the following step 7306 .

Dictionary 7018 is retrieved for "26" in the input string to take out dictionary information indicating that "26" is a "number" ( 7306 ). Then morpheme processing information indicates that "number, number" is a morphemic feature, that is, it is a sequence of numbers and is treated as a grouped cardinal number ( 7308 ). It is now judged whether or not the group receiving the dictionary information has been provided with the morpheme processing information at step 7308 ( 7130 ). If it has been provided, the flowchart proceeds to step 7314 to apply further processing based on the local analysis rule, and no designated group is recorded in table 7126 ( 7312 ); the flowchart then returns to step 7302 . Thus, step 7314 advances to "26" because this number is provided with the morpheme processing information. The processing at step 7314 is performed according to the operation of the flow chart shown in Fig. 55B.

First, an initial value "0" is set in a counter n , which counts the number of adjustment reference units when the reference units are retrieved in the adjustment retrieving section 7112 ( 7410 ). Further, since the reference unit completed by dictionary retrieval in section 7010 is set as "p" , the adjustment table 7128 for the p + n _th (n = 0) reference unit, ie, "26", is set by the section 7112 accessed ( 7412 ). Since "26" has been provided with the morphemic processing information which indicates that it is an atomic number at step 7308 , and since these compilations "each with an atomic number" at the top end and after the second expression, the compilation in the Adaptation table 7128 are present (see FIG. 57), the reference unit "26" is equal to the information in the adaptation table 7128 and consequently adapted with respect to it. In this case, an adjustment for Ms-Me is performed while the second adjusted compilation is set as "Ms", with the last data of the combination having the "ordinal number" at its top as "me" in the adjustment table 7128 .

Based on the fit result in table 7128 at a (p + n _te (n = 0) dictionary reference unit, the fit state is judged ( 7414 ) and if it is determined that it is fit, the flowchart advances to step 7416 while if it is determined that it is not matched, the flowchart proceeds to step 7424 .

When it is judged that to be adjusted, "1" is set in the counter n to perform the separation for the dictionary separation unit at p + 1 _te (n = 1) in the input string. The separation is performed in the same manner as in step 7302 . Dictionary 7018 is retrieved for "Jan.," in the input string, since this has been deposited as the dictionary reference unit closest to "26" for processing to provide the morpheme processing information ( 7420, 7422 ). This processing is performed in the same manner as in steps 7306 and 7308 .

By repeating the operations from step 7412 to step 7422 , the flowchart up to "Jan 26, '80 he" is performed closed. However, since "he" does not match the adjustment with respect to the adjustment table 7128 in step 4712 , the flowchart advances to step 7424 in step 7415 . This means that although the dates up to 26 Jan. '80 "with" Cardinal number, month, year "in the adjustment table 7128 have been adapted, they are not adapted for" 26 Jan.,' 80 he ".

Furthermore, when the sentence on the input string, for example "26. Jan., '80'" has ended, ie there is no next elimination for the next dictionary reference unit, step 7418 advances to step 7424 in the flowchart at step 7418 . In the event that it is determined that at step 7414 is not matched, it is now judged whether the content of the counter n is not more than 1 or not ( 7424 ), and if it is not more than 1, this is considered one only reference unit recorded in table 7126 ( 7434 ).

If it is not less than 1, an adjustment is made by taking the p + n (n = 3), ie "he" in "6 Jan., '80 he" is "EOS", thereby ending the compilation is displayed ( 7426, 7428 ). If it is not matched, step 7434 is advanced to in the flow chart. When adjusted, "Jan. 26, '80", which is p - ( p + n -1) for the dictionary reference unit, is collectively compiled according to the compilation result, which corresponds to the compilation Ms in the adaptation table 7128 , and that Result is recorded in table 7126 ( 7430 ). Then it is noted that the reference units at the (p + n-1) th unit have ended and (p + n ... 1 ) is reset to "p" ( 7432 ).

The eighth embodiment according to the invention will now be explained. In Fig. 63, the entire structure of the eighth embodiment is shown of a speech analyzer according to the invention, which / Japanese is applied to an automatic transmission device for English.

Fig. 63 shows an input section 8801 , an English text 8002 , a pre-editing section 8003 , a section 8004 for a morphological analysis, a section I 8005 for a syntactic analysis, a section Ii 8006 for a syntactical sentence analysis, an operating display section 8007 , a word dictionary 8008 , a rules file 8009 , a control section 8010 , a structure converting section 8011 , a section 8012 for generating a translated sentence, a redesigning section 8013 , an output section 8014 and a Japanese sentence 8015 . As shown in FIG. 63, this translation device has the input section 8001 via which the English text 8002 to be translated into Japanese is input. The input section 8001 may include , for example, a keypad with character keys such as alphanumeric and function keys, an optical character reader (OCR) for reading the English text, and / or a file storage device for reading the English text, which is recorded in a storage medium such as a magnetic disk .

The English text entered in section 8001 is read into the pre-editing section 8003 , whereby a pre-treatment for the translation is carried out. In this case, sentence recognition and processing of unknown words are mainly performed. This acts as part of the morphological analysis.

The pre-edited English data, along with the information obtained from pre-edited, is provided to section 8004 . Section 8004 divides the data by referring to word dictionary 8008 , decomposes the English morphemes, performs various types of merging, such as processing unknown words, expressing a proper name, time and numbers, and processing for the entire sentence, like an additional question (tag question) and an apposition detection. The rules for a morphological analysis are contained in the rules file 8009 .

The English data, after being subjected to the morphological analysis, together with the dictionary information obtained by the morphological analysis, are given to section I 8005 . Section 1 8005 is a functional section that breaks down the surface layer structure for the sentences by applying a grammatical rule to the English data and finds all structural possibilities.

The English data, after being subjected to the syntactical analysis in section I 8005 , together with the corresponding information, are submitted to section II 8006 , where a solution from the result of the syntactical analysis with regard to the surface layer by the syntactical analysis I is selected using a structural description. Consequently, a plausible so-called parsing or analysis tree of the English description is prepared to form its structure. These so-called parsing or segmentation rules are also contained in the rules file 8009 .

The English data which has been subjected to the parsing is transmitted to a structure converting section 8011 as the data for the corresponding analysis tree. Section 8011 prepares a corresponding Japanese structure tree from the structure tree, which is an intermediate structure of the English sentence, in order to convert it into a structure underlying the Japanese, from which the Japanese sentence can then be easily translated.

The data for the structure tree, which indicates the underlying Japanese structure that has been transformed in this way, is supplied to a translation section 8012 , in which the translated sentence is generated. This is a functional section for creating a Japanese sentence from the structure of the structure tree corresponding to the Japanese sentence.

The Japanese sentence data thus translated, that is, the data for the Japanese sentence, is supplied to the pre-editing section 8013 , which modifies the translated data to retrieve the memory 8008 , using the information used in the translation to make the Japanese sentence more natural perfect. The data for the Japanese sentence is transferred to the output section 8014 and output as the translated Japanese sentence 8015 from the output section 8014 . The output section 8014 has a printer, a display, and / or a file storage device such as a magnetic disk.

The flow of a series of translation operations is controlled by control section 8010 , which controls control for the entire facility. In the illustrated embodiment, the word dictionary 8008 contains dictionary data for English and Japanese words, and various information, such as a connecting relationship, that is to say a concurrent relationship, meanings, singular and plural forms, a language part, etc., is also specified for the vocabulary. The rule file 8009 also contains rule data for morphological analysis and for an English sentence breakdown.

The control section 8010 is connected to the operation display section 8007 , which has operation keys for issuing various commands from an operator to the device, such as a translation command key, a cursor key, etc., a display or a display which visually shows the inputted English sentence text Japanese sentence as a result of translation, intermediate data, such as dictionary information and various commands to the operator. It can be designed so that most of these operation display functions are present in the keypad if it is provided in the input section 8001 or on the display if it is provided in the output section 8014 .

The embodiment of the invention relates to the automatic translation device as described above, and is used so that when a derived word is contained in the English text 8002 , a grammatical characteristic, a semantic characteristic equivalent (Japanese), etc. depending on the conditions for the derived word, which has been recognized by an addition, for example, to thereby increase the reliability of the result of the decomposition or the translation. An affix dictionary is used to evaluate dictionary information for unknown words in the morphological analysis. Three types of processing are required as processing modes, ie processing for the prefix or prefix, a prefix for the suffix or suffix and the evaluated processing by the suffix. However, the types of data essentially contain two types, that is, prefix and suffix rating data.

First, the prefix and suffix rating data explained in detail.

(1) Prefix rating data

If the prefix or prefix part of one is not in the dictionary registered word with the one described below (Part) matches and the remainder in the dictionary exists, the word will match the dictionary data treated for its root. With the dictionary data is it is possible to add an internal characteristic to the series of the original add internal feature and a Japanese Add suffix to the original translation word. For the word "electrochemical", for example can the "electrochemical" input in the dictionary information Preservation table according to the dictionary data through the prefix input "electro" and the dictionary input "chemically" as shown below is:

Prefix Entry Dictionary Entry Electrochemically Entered Word Dictionary Information
Table input electrochemical electrochemical

The input in the table then takes over all the dictionary Information to which the word roots correspond.

(2) Suffix evaluation data

If the suffix or suffix part of one is not in the dictionary registered word with the one described below (Part) matches and the rest of the part in the dictionary this word is registered by adding new Dictionary data is created according to the information, which is described in the suffix dictionary. In this Case becomes the first (Japanese) equivalent in the dictionary data, that correspond to the root part of the word taken out and for the (Japanese) equivalent in the new dictionary data used.

For example, for the entered word "controler"  an input in the table "controler" based on the Suffix input "- (e) r" and the dictionary input "control" registered.

Suffix input dictionary input (verb) - (e) rcontrol main word verb input word dictionary information
Table entry controlercontroler noun

Now the outline for processing a derived one and explains the processing of an unknown word.

• (1) If for a word not registered in the dictionary a prefix or suffix at the beginning or end of a Word and the rest of the word in the dictionary is registered, become an English part a language information, an internal feature and the Japanese Equivalent based on the dictionary and the Affix information composed.
• (2) Prefix and suffix are listed as one film system and can be edited independently of a program.
• (3) First the possibility for the prefix is tried and if this is not possible, the possibility for tried the suffix. If both are included, no attempt made.
• (4) For the word that failed the trial evaluation is processing for the end part as one unknown word performed.

In particular, an eighth embodiment according to the invention will now be explained with reference to a block diagram in FIG. 58. In Fig. 58 are shown, an input processing section 8020, an input Aufteilabschnitt 8021, a definition table 8022, a dictionary retrieval section 8026, a derivation processing section 8023, a reference dictionary 8024 and a preservative table 8025 for a dictionary information. First, an English sentence is written in the input processing section from an input device having a template input file or keyboard, an ORC, and so on. The dictionary reference unit is then divided into the section provided for this purpose, reference being made to the delimitation table, and if this cannot be ended, a dictionary attempt is carried out using the reference dictionary. If there is an input as a result of the attempt, the test result is recorded in the dictionary information preservation table, while processing for a derived word is performed when there is no input.

Fig. 59 is a block diagram for explaining an embodiment of the discharge processing by a prefix. Fig. 49 shows an adaptation section 8030 between the upper part and the prefix dictionary, a prefix dictionary 8031 , a section 8030 for dictionary retrieval except for the prefix part, an adaptation section 8033 for the speech part in the prefix dictionary and the speech part of the input, a dictionary information preparation section 8034 by the prefix rating and a derivation processing section 8035 by the prefix.

Fig. 60 is a block diagram for explaining an embodiment of the discharge processing by a suffix. Shown in Fig. 60 are an adaptation section 8040 between the end part and the suffix dictionary, a suffix dictionary 8041 , a processing section 8042 for completing a (completely unknown) unregistered word, a dictionary retrieving section 8043 for the dictionary with respect to the part other than the suffix Part, a matching section 8044 between the language part of the root and the input language part in the suffix, a dictionary information preprocessing processing section 8045 by suffix evaluation, a dictionary retrieving section 8046 for performing dictionary retrieval by adding the root change in the suffix to the Part except for the suffix part and a non-registered word processing section 8074 by means of a suffix evaluation. First, an adjustment is made between the end part and the suffix dictionary. If it is not matched, the word is processed as a fully (not unknown) unregistered word, while if it is matched the dictionary is retrieved for the part except for the suffix part. If there is no input as a result of dictionary retrieval, dictionary retrieval is performed while the root change in the suffix dictionary is added in part except for the suffix part. If there is no input in the result, processing for an unregistered word is performed using the suffix evaluation. On the other hand, while if there is an input, an adjustment between the speech part of the input and the speech part of the root is performed in the suffix dictionary in the same way as in the case where the input is the result of dictionary retrieval for the part except for the suffix part is present. If it is matched, dictionary information preparation processing is performed by a suffix evaluation, while if it is not adapted, processing for an unmatched word is performed by a suffix evaluation.

Fig. 61 is a block diagram showing details of the whole structure obtained by assembling the parts of Figs. 58 to 60; FIG. 652 shows the details for the processing section 8052 shown in FIG. 61 for the fully (hardly unknown) unregistered word. The fully unregistered word processing section 8042 shown in Fig. 61 has a processing section 8050 for a proven information preparation for a noun and a corresponding section 8051 for a verb. However, since each of the parts in Figs. 61 and 62 has already been described in detail, they need not be explained again.

A ninth embodiment according to the invention will now be described. Here, 64 of the entire structure of the ninth embodiment is shown in Fig. According to the invention, which is applied to an automatic transmission device for English-Japanese. This embodiment has an input section 9010 via which an English text 9012 to be translated into Japanese is input. The input section 9010 may include a keypad with character keys such as alphanumeric or function keys, an optical character reader (ORC) for reading the English text recorded on paper, and / or a file storage device for reading the English text, which is stored on a storage medium such as a magnetic disk , is recorded.

The English text input from the section 9010 is read into a pre-editing section 9014 in which the translation pre-treatment is performed. In this case, sentence recognition and processing of unknown words are mainly performed. This then acts as part of a morphological analysis.

The pre-edited English data, along with the information obtained from the pre-editing, is transferred to a section 9016 for morphological analysis. Section 9016 divides the phrase for a word dictionary query 9018 by dissecting the English morphemes, performs various arrangements such as unknown word processing, proper name processing, expression for time, number, etc., and also performs one Processing for the entire sentence such as an additional question (tag question) and an apposition detection by. The rules for a morphological analysis are contained in a rules file 9036 .

The English data which has been subjected to the morphological analysis, together with the dictionary information obtained from the morphological analysis, are transmitted to a section I 9020 for syntactic analysis. Section I 9020 in this embodiment is a functional section that breaks down the surface layer structure from bottom to top and right to left for the sentence by applying the CFG rule to the English data and finds all structural possibilities.

The English data, after having been parsed in section I 9020 , is submitted along with the parsed information to section II 9022 for parsing. The section selects a solution to the result of the syntactic analysis with respect to the surface layer structure in section I by applying a structural description. In this way, a plausible parsing or analysis tree is prepared for an English sentence in order to form its structure. These so-called parsing rules are also stored in the rule file 9036 . The English data which has been subjected to the parsing analysis is transmitted to the structure converting section 9024 as the data for the so-called parsing tree. In section 9024 , the so-called parsing tree for the corresponding Japanese sentence is prepared for a structural tree, which is an intermediate structure of an English sentence and is transformed into a structure underlying the Japanese, from which a Japanese sentence can then be easily translated. The data for the structural tree, which indicates the underlying Japanese structure that has been converted in this way, is provided to a translation section 9026 , in which the translated sentence is generated. This is a function to create a Japanese sentence from a tree structure of the Japanese structure tree. First, a sentence structure is created by changing the structure tree by swapping the order so as to conform to the structure of Japanese, and then morpheme creation is performed to convert a translated sentence from top to bottom and left to right in the sentence structure tree.

The data for the Japanese structure tree thus created is translation data that is supplied to a redesigning section 9030 , in which the translation data is modified by retrieving the dictionary 9018 with the help of the information used in the translation processing to make it more natural to complete Japanese sentence. The data for the Japanese sentence are transferred to an output section 9032 and then output as the translated Japanese sentence 9034 from the output section 9032 made. The output section 9032 has, for example, a printer, a display and / or a file storage device, such as a magnetic disk. The flow of a series of translation operations is controlled by a control section 9038 which regulates control for the entire facility. The dictionary files 9018 stores dictionary data for English and Japanese words, and the rules file 9036 stores data for morphological and syntactic analysis in this embodiment.

The control unit 9038 is connected to an operation display section 9040 which has operation keys for issuing various commands from an operator to the device, such as a translation command key or a cursor key, and a display or a display device which visually displays an inputted English text, a Japanese sentence as the translation result, intermediate data such as dictionary information and various commands for the operator. It may be configured so that most of the operation display functions are included in a keypad when it is located on the input section 9010 or on a display when it is located on the output section 9032 . In section I for a syntactic analysis, the cfg rule is used for the English sentence from top to bottom and from right to left for the English data after a morphological analysis in order to derive all possible solutions for the sentence structure. The solutions are generally to be understood in the form of a structure tree. This shows a relationship for words or groups contained in each of the sentences that are related to each other in a subordinate or coexisting relationship, such as a modified relationship or a case relationship, a subordinate relationship to one another, such as between parents, child and grandchildren, etc. related. Each of the words or the group is located at the node of the structure tree.

In this embodiment, the syntactical analysis Features that respect shape and vocabularies one sentence differentiated to the collective compilation to judge in terms of sentence structure. The Compilation with regard to the sentence structure is here referred to as "unit" and "block".

The "unity" is a word or a group of words, which is the smallest unit for the translation process form which is identical to a word in the syntactic Analysis is treated, and the dictionary information  for each of the legal elements contained therein is not used.

A "block" is a structural merger in which the syntactic analysis preferably for their interior is carried out with respect to the outer part and which in an equivalent way as a unit with respect to them outer part is treated. For example, it can be a Part of a sentence, a phrase etc. as well as what the intermediate symbols, corresponds to which is used in the cfg grammar have been. It can also have a nest structure, i.e. H. a Block can also contain another block. Furthermore, the term block also includes a sentence, a paragraph, and whole Include sentences, each of which is considered a block can be. The processing, which of the partial Syntactic analysis gives preference here as one partial dissection or analysis "designated". Hereby can be uneconomical structural solutions that have been described above, are degraded to the analysis efficiency to improve, and to make a more believable Get analysis result.

For the block, two features become in this embodiment Are defined. One of them is a symbol in the cfg rule, referred to as "target" in the present specification will put which one as the result of the analysis is that with each of the constitutional elements inside of the block has been performed, d. H. a symbol that describes the structure or property of the block. The other is a symbol in the cfg rule, as one "Role" is referred to, which is carried by the block, if the parsing on the outside of the block for the sentence, phrase or part of the sentence is carried out, in which the block is contained. I.e. it is a symbol that is the relationship or role of the block to the others describes.  

For example, in the case of an English sentence I "White House isn't white" is the goal of a sentence, and that Role is a noun (a part of a sentence). Although the goal and the role is generally identical in most cases are, as in this case, sometimes different from each other.

For the embodiment shown in FIG. 62, if the structural arrangement of the input English sentence is recognized as a block, the functional sections to evaluate their purpose and role are summarized in the structure as shown in FIG. 65. As can be seen from Fig. 65, the structural arrangement of the English sentence types which have been pre-edited in section 9014 is distinguished in section 9016 for morphological analysis using the word dictionary 9018 and the rule file 9036 .

The word dictionary 9018 stores dictionary information for English words and phrases. For example, as shown in Fig. 68, inputs are formed with variations for each word in this embodiment, and all information is developed. For example, a number of pieces of speech information can be provided for the speech part information, as shown in FIG . The way to build dictionary 9018 is not limited to this example only.

The rule file 9036 stores this data for the upper state indicating the upper end of the block, the end state indicating the end, and block preparation information for creating the block with the target and the role in the form of a table. An example of this is shown in Fig. 69. For example, a block is started by "conjunction" and ends at the end of the sentence. As a result, a block is formed that begins with "," from the top, which immediately precedes its conjunction, and the goal is a clause, while the role is a sentence. Furthermore, a further block is formed, starting from the conjunction up to the end of the sentence, in which both the goal and the role are part of a sentence or a clause.

Furthermore, a block is started by ", relative pronoun" and is ended by "," or at the end of the sentence. As in this case, the possibility for a number of final conditions for an upper condition is allowed. In the event that the block is ended by ",", a point (cluster) forms "," which precedes the relative pronoun until the next occurrence ",", a block in which the target is a sentence or a sentence Clause and the role is an adverb or an adjective. That is, this means that the entity acts as an adverb or an adjective clause. In the case of an ending at the end of the sentence, an educated person "that precedes the relative pronoun at the end of the sentence, a block in which the goal is a clause and the role is an adverb or adjective. This is according to the conditions for forming a group, clause or sentence that appear in common modern English sentences. In the figure , the symbol "" represents a space. In the morphological analysis section 9016 , the English text input from the pre-editing 9014 is first divided into sentences as the translation units. In this case, incorrect spelling or unregistered words are detected. The dictionary 9018 is retrieved from the respective sentence units and the dictionary information for each of the components is fetched. Different arrangement modes are carried out according to this dictionary information.

Figure 66 shows a flow diagram for the collective assembly of a block performed in section 9016 for morphological analysis. First, a position pointer indicating the reading position for an English sentence is set at the upper end ( 9100 ). The upper end does not mean the word at the upper end, but the upper (imaginary) end of the sentence just preceding. Word extract 9101 is performed at this point. As shown in FIG. 67, in operation 9101, a word is taken out by advancing the position by one ( 9111 ) if it is not at the end of the sentence ( 9110 ), and dictionary 9018 is retrieved for the word ( 9112 ) to write out the word information ( 9113 ).

When the word information is extracted in the process 9101 , reference is made to the table 9036 for the conditions at the beginning and end of the block to judge whether there is anything that matches the above condition or not ( 9102 ) . In this way, steps 9101 and 9102 are repeated until the word matching the above condition is found.

If the above condition is met, the next word and subsequent words are taken out by the requested number and reference is made to the match with the top state of the block ( 9104 ). In this case, the dictionary is retrieved for each of the words if necessary. The position pointer is not advanced.

If compared to the top state of the block at step 9104 , then a word that matches the end of block state with respect to the top state is retrieved ( 9105 ). Steps 9104 through 9106 are repeated until the word that matches the end condition is found. If a word matches the end condition ( 9106 ), a so-called cluster that includes the word is recognized as a block and the block is written ( 9107 ). In particular, a block is prepared to judge whether the block preparation condition is satisfied at the point where the end condition is satisfied first. Regarding the block preparation information table 9036 , the position for the word indicated by the pointer at the point where its advancement in processing 9103 is stopped is set as the top position for the block and the position of the Word that meets the ending condition that appears first after that is set as the ending position for the block. At the same time, the goal and role of the block are written.

If, as a result of such block detection, "... Conjunction..." for example, in the English sentence, as shown in Fig. 70, the entity from the top of the sentence at the portion preceding "," is recognized as a block, while the entity from ", conjunction" is recognized at the end of the sentence is recognized as another block. In the figure , the inside of [] indicates a block. In this block, both the goal and the role are sentences. Furthermore, a structure from the word after the conjunction at the end of the sentence forms a further block in which both the goal and the role are sentences. On the other hand, the entity from the conjunction at the end of the sentence can be defined as a block. In this case the goal is a phrase or a clause and the role is an adverb.

The block can also be set as it is by the Position off that does not contain a ",". Furthermore, the Punctuation and Ä. of the subject for grammatical Analysis as the information to be excluded that exists in the block. If in the same way ". . . Relative pronoun . . ." is present, a "relative pronoun..." be recognized as a block. In this block the goal is a clause or a phrase and the role is an adverb or an adjective.  

The block can of course also be designed in a nest structure. For example, if the English sentence has a structure such as "(beginning of sentence"..., Conjunctions,..., Relative pronouns... ") End of the sentence, as shown in Fig. 71 for example, then the construct of ", Conjunction" until the end of the sentence a block BL1-BL1, in which ", relative pronoun. . . "is contained as another block BL2-BL2.

In this way, the morphological analysis section 9016 distinguishes the feature of the sentence in terms of shape and vocabulary to distinguish the structural composition as a block. Furthermore, in addition to such a block recognition, section 9016 carries out various processing operations, such as an expression for a proper name, a derived, unknown word, an abbreviated word, a numerical, a time, a hyphenated word, an apostrophe (′) and an apposition assessment and processing for an additional question (tag question) in order to prepare morphological analysis data.

The English sentence, which has been subjected to a morphological analysis in this way, is transmitted together with the analyzed information to the syntactical analysis section I 9020 . Fig. 72 shows an example of the output data. Fig. 72, the result is that an English sentence I "White House is'nt white" from the input section 9010 is input, and is morphologically analyzed in the section 9016th Block 1 is started at word position # 4 and ended at position # 10, in which case both the target and the role are optional. In this way, block 2 is started at position # 5 and ended at position # 6, the target being a noun group, while the role is a proper name. That is, the block "White House is'nt white." contains white house in a different block than a nest. In a block, ie a smaller "White House" block, each of the components inside functions as a proper name, while it has a position as a word clause relative to the exterior, ie "is'nt white." . "White House" can be treated as a unit.

Together with such block information, word information which has been retrieved from the word dictionary 9018 is added and output from the morphological analysis section 9016 to the syntactical analysis section I 9020 . Section I 9020 analyzes the surface layer structure of the English sentence by applying a context-free grammar rule, which is stored in the rule file 9036 , in order to find out all possible structural trees. In this case, if a block is included, the partial grammatical analysis described above is performed while referring to the local analysis. As a result, the efficiency and accuracy of the respective analysis can be improved.

In particular, the block inclusion relationship is determined by the position information prepared for the block. Then the innermost block analyzed. The block that comes with this grammatical Analysis has ended as a unit considered, and no further processing in its Performed inside. This way the area of grammatical analysis gradually to the outside Blocks enlarged. Finally the whole sentence is analyzed. The respective analysis is based on the cfg Usually from top to bottom and right to top performed on the left in the English sentence. The grammatical Analyze is done in a manner where all possibilities remain, which are given by the grammar rules allowed are.

An example of such an analysis processing flow is shown in FIG . First, based on the English data supplied to Section I 9020 , all structural compositions for a set are recognized as blocks and the goal and role are evaluated ( 9120 ). The route to a compilation is as shown in Fig. 70. Then, if there is no block in such an arrangement ( 9121 ), the sentence is analyzed ( 9125 ) and only a sentence collectively composed as a symbol is selected and the analysis for the sentence is ended ( 9126 ). Since processing steps 9125 and 9126 are included in processing steps 9121 to 9124 when the processing system treats the whole sentence as a block, they are not unnecessary.

If there is a block, the innermost block is analyzed first ( 9122 ). In the example shown in Fig. 71, the inside of the block BL2-BL2 is analyzed. Although various solutions are generally obtained by grammatical parsing, a solution in which the block is collectively assembled as a cfg symbol and which matches the target of the block, selects these solutions ( 9123 ). In this case, all that is arranged in a symbol is selected for the block with regard to the optional target. Everything that has been selected in this way is then treated as a single arrangement that has the role for the block ( 9124 ). In the block with the optional role, a role of the symbol which is selected in processing 9123 is defined as the role. The processing operations 9121 to 9124 are then repeated one after the other.

In this way, in the example of Fig. 71, the inside of the block BL2-BL2 is first grammatically analyzed and then the inside of the block BL1-BL1. In this case, block BL2-BL2 is treated in the same way as a single word and each of the components contained therein is not analyzed grammatically.

If the data defining the structural composition and the subordinate relationship are held in this way, they are delivered to the syntactical analysis section II 9022 . The data can then be easily recognized in the form of a structure tree, as described above. The data is then further transformed into the structure of the Japanese sentence in the structure converting section 9024 and in the translation section 9026 ; a translated sentence is then generated for each of the nodes contained therein. The node processing in the structure construction is carried out in this way from top to bottom and from left to right.

The set so generated, translated is then subjected to post-processing in the Nachredigierabschnitt 9030, is visually displayed on the display section 9040 and Operation for example, printed out as a Japanese sentence 9034 in the output section 9032nd

In this way, according to this embodiment, the feature of the English sentence with regard to the form and the Vocabulary differentiated, so the structural composition to distinguish as a block. For the block become a goal, which can be the analysis result, and the structural role with which the block is outward acts, evaluates. Then the surface layer structure of the English sentence analyzed by a context-free Grammar rule is applied to all possible Find out structure trees. This can increase the number of uneconomical solutions, the efficiency at grammatical analysis are improved as well as a more reliable analysis result can be created.

So there are different patterns for one to add Expression, and it is difficult to do this at grammatical Analysis, especially with context-free analysis recognizable as it is in view of the foregoing  is generally difficult to post-apposition detection performing the analysis is an ambiguous translation inevitable. Even if a rule has been prepared with whichever they are, it would be a risk that no appositional expression to be added as an identical one Case is detected, or the number of possible combinations becomes significant. That said, it will be an expensive one local analysis between the parts contained in the appositional expression are included, and the other parts carried out.

In view of this, in the present embodiment alleviated the load in processing in the analysis step by the appositional expression by the Characteristic of the sentence in terms of form or semantic Characteristics of words is recognized. An assessment of the Apposition is done by using the next pattern as a block is recognized.

For the English sentence structure "∼, relative pronouns, ∼.∼" the relative pronoun is recognized in that the language part code for the word with a special code, for example "R" is provided. In this case, the inside that comes with "," is considered as a block provided that he's not dealing with a block or a unit overlaps what has been shown in the pre-edit and it is not "and" or "or" in the part after second "," contains. For the English sentence structure "∼, Relative pronoun ∼. "Becomes the interior, that of", "and", " is considered as a block. The point can be another symbol that is used for the end of the sentence becomes.

The dictionary 9018 is designed to carry out such an apposition evaluation in order to store the meaning information for words. The meaning information represents the difference for the article, location, person, etc., as shown in Fig. 74. Also for the block preparation condition, the table 9036 is constructed as shown in Fig. 75, so that the top of the block is recognized as the top condition by "proper name (person), noun (person)" and the top End of the block is recognized by "proper name (person), article noun (person)". It is thus possible to evaluate the appositional expression in terms of the morphological and semantic characteristics without performing a grammatical analysis and to perform a grammatical analysis according to the appositions evaluation in the example shown in FIG. 64 for other processing operations.

Incidentally, in the English sentence it is these groups that carry a very special information and only limited be used. If analyzed in the same way will be, as is done for the usual groups analyzed them for a completely different type of sentence, and it is difficult the original nature of the sentence to get by analyzing. Furthermore, this leads to large ones Loss.

For example, "let's" or "let us" immediately after the punctuation etc. as a command set with the causative Verb "let" parses and it should be as a group be analyzed, which the inviting character "let" ("let's" and the like). "Let" can also be used in various ways be used as a transitive verb and even as a noun in the meaning "rental" and is not on the use of friendly invitations or Prompts and not limited to use as an auxiliary verb. Consequently, the analysis has to be done for the respective possibilities be carried out, reducing the efficiency is reduced. Furthermore, it is difficult to use as Invite to result from the result of the analysis as it is except for the sentence structure, no difference  between use as a causation and as an invitation, and it's difficult just because of the sentence structure to distinguish.

The uneconomical loss in the course of an analysis can can be reduced by "let's" or "let us" ("let us ") immediately after the punctuation of the object the analysis is excluded. By separating from the basic use of the word, d. H. from the causative Semantic analysis can easily be used be performed.

If "please", "let′s" or "let us" appears at the beginning of the block, there is a flag for the Block information is used, and this is for each of the Cases not given in information for the unit. For example, the English phrase "let’s go to school." ("Let's go to school") processed as <go to school < ("go to school") <to which let's ("let's") appended is <.

In order to perform such "let" processing, a let information processing section 9200 is arranged between the morphological analysis section 9016 and the syntactical analysis section I 9020 in a modification of this embodiment. Fig. 77 shows the parts in question. In this figure , the same elements as those shown in Fig. 64 are given the same reference numerals.

Dictionary 9018 is also designed to store let information for the word. As shown in Fig. 78, let information creates "0" for ordinary words, "1" for "let's" and "let us" and "2" for "please".

The section 9200 for let information processing has the task of taking the result of the morphological analysis together with the input English sentence from section 9016 and adding the let information as the additional information to the information of a word during the analysis as shown in FIG . is shown 79th In this case, a block is arranged for the sentence. In the example shown in the figure, block is 0 (start: 1, end: 10, goal: sentence, role: sentence). That is, the block in this example includes a clause, a group, etc., in addition to a sentence. In this case, the term for the block also includes a paragraph and the entire sentence, each of which can be viewed as a block. Further, a "transitive verb root (which is appended with 's)" is written as a language part to "let's" for information of the word, and the let information is "1".

As shown in Fig. 81, processing to prepare a block of sentence 9300 is performed before starting the collective assembly of the block into the input English sentence. The subsequent processing operations are the same as the flowcharts shown in Fig. 66. For example, in the English sentence, I said "let’s go to school", (I said, "let's go to school") block 0 is formed (start: beginning of the sentence, end: end of the sentence, role : Sentence, goal: sentence).

As shown in Fig. 81, in the syntactical analysis section I 9020, each of the structural compositions is recognized as a block based on the English data supplied to it, and its goal and role are evaluated accordingly ( 9120 ). If the block does not exist in the array, ( 9121 ) the analysis is ended. If there are blocks in the input set, the innermost block is analyzed first ( 9122 ). Although various solutions are generally obtained by analyzing, only the solution collectively arranged as a cfg symbol is selected from among them ( 9123 ). The subsequent processing operations are the same as those in FIG. 23.

Such processing of "let" information is performed in section 9200 according to the processing flows shown in Figs. 83A and 83B. First, a pointer is placed on the top block ( 9330 ) to check the word located at the top or beginning of the block ( 9331 ). If the let information is "0", the pointer is incremented ( 9339 ) to get the next word over.

If the let information is not "0", the previous dictionary reference unit is checked ( 9322 ). If it is not punctuation, or if the pointer does not indicate the top or beginning, the pointer is incremented ( 9339 ) to proceed to the next word. If the previous dictionary reference unit is a punctuation in the check, or if the pointer indicates the top or the beginning, the innermost layer block containing the word is marked ( 9333 ).

If the let information is then "1" ( 9334 ), since this is "let's" or "let us" immediately after the punctuation, the role of the marked block is recognized as (invitation sentence ) ( 9336 ). If the information is "2" because it is "please", the role of the highlighted block is recognized as a "prompt set" ( 9335 ). Then the destination of the highlighted block is recognized as an instruction set ( 9337 ) and the word information indicated by the pointer is deleted ( 9338 ). Then the pointer is advanced ( 9339 ) to proceed to the next word. The procedure is carried out up to the word at the end position ( 9340 ).

Fig. 80 shows examples of the result of analysis has been carried out with which such a let-information processing in the example of the aforementioned input sentence: I said, "let's go to school." (I said, "let's go to school.") When section 9200 adds the let information to the information for the word, it removes the information related to the let information from the table, and the block information becomes as "Imperative sentence" described as the goal and as an "invitation sentence" as the role, as shown in Fig. 80.

If dictionary 9018 is retrieved in the treatment of hyphenated words in the English sentence as the entirety of a number of hyphen-connected words, if their inputs are present in dictionary 9018 , the method is successfully performed. If for the hyphenated word that is not registered in dictionary 9018 , the whole part is treated as an unknown word, e.g. B. an adjective, since the dictionary information cannot be used for each of the words connected with a hyphen, it cannot be translated. Furthermore, if there is input for the information for each of the constituent parts of the hyphenated words in the dictionary 9018 , they cannot be neglected.

If the analysis is also carried out by: are broken down into their respective components, Art a connection to hyphenated words extremely versatile.

To solve this difficulty, all of the hyphenated words are analyzed as an adjective in the sentence, and a corresponding analysis is performed only for the inner part of the hyphenated words, using the components of the hyphenated words, and the result of this analysis is then combined. This also enables analysis for hyphenated words by using the information for each of the components. That is, for hyphenated words that are not registered in dictionary 9018 , the whole part is treated in the same way as an adjective. Reference is made to the nouns connected by the hyphen in the dictionary, and the corresponding analysis is carried out in a closed form only for the inside of the words connected by the hyphen.

That is, if the hyphenated word is a word not registered in the dictionary 9018 , block information is given that the whole part is regarded as a block, and the dictionary reference is performed for each of the constituents for the inside of the block by the respective unit information in which the hyphen is not included. For the word not registered in the reference dictionary, final evaluation processing is performed together with an unknown word processing as a kind of processing operation for unknown words.

Such processing for hyphenated words can be performed in the example of the structure shown in FIG. 64. In this case, the position of the word in the sentence is expressed not by the number attached to the word but by the number of characters from the beginning of the sentence, ie the number of characters.

In Fig. 84 an example of the processing of hyphenated words is shown, which is performed in the morphological analysis section 9016th For the English phrase entered, for example, "The anti-war attitude is her open-door policy". The cursor is incremented to take out a word ( 9135 ) and then performs dictionary retrieval ( 9353 ). In this case, the hyphen is not used as a delimitation for the word. If the input is present ( 9353 ), the word information is written out ( 9359 ).

This is then repeated until the end of the sentence.

In the event that the input does not exist as a result of dictionary reference 9352 , if there is no hyphenated word ( 9354 ), word information is written out ( 9359 ), while if there is a hyphenated word , a block is written with the hyphen ( 9355 ). In the hyphenated block, the starting position is the starting position for the hyphenated word and the ending position is the ending position for the hyphenated word. The goal is optional and the role is adjective / noun. The hyphen is then released to take out each of the constituent words ( 9356 ) and the respective constituent words are retrieved from memory ( 9357 ). The word information obtained as the result of dictionary retrieval ( 9358 ) is written. When the word information is written out at steps 9359 and 9358 , in the case of a word not registered in the dictionary, it is written out as a speech part = a word not registered in the dictionary.

Fig. 85 shows examples of the block information and the word information of the English block which is collectively assembled into one block as a result of processing the input sentence example. In this example, hyphenated words "anti-war" are registered in dictionary 9018 , and the words "open-door" are not registered in the dictionary. Thus, the input for the hyphenated word "anti-war" is written out as the information for the word. For the hyphenated words "open-door" they are broken down as "open" and "door" and written out as the information for the words, and block 1 (start: 30, end: 38, goal: optional, role: adjective / Noun) is written out as the information for the block. Although the form of the English question (tag question) is extremely limited, its processing in the usual analysis procedure is very complicated. Furthermore, it is not easy to determine the verb to which a so-called additional question relates. After recognizing that it is a so-called additional question based on the feature of the sentence in terms of its form, it is treated as an additional question relative to the associated structural composition, whereby the verb affected by the additional question can be specified . That is, the part of the supplementary question in the English sentence is found out as a structural pattern and the analysis is carried out, the part of the supplementary question being regarded as pure information with a certain type of associated structural composition.

In the present embodiment, a unit or described a block in the form of a symbol (a starting point display, which indicates that this is a unit or Block end point is). In the morphological analysis the entered sentence text shaped, in which also the recognition is carried out for the block. In the present embodiment "a quotation mark" is referred to as "Q" and a bracket is referred to as "P". For example this is determined as follows:

'. . . .'By (Q '....)', "...." by (Q "....)", (....) by ((P....)), <. . . . <by <(P....) <, {. . . .} by {(P....)} and [. . . .] by [(P....)] or

The block detection is carried out in the same way.

The start and end symbols of the block are only under applied to the context where the block is opened by these symbols 39104 00070 552 001000280000000200012000285913899300040 0002003733674 00004 38985 or is closed. The part immediately before  Start signal and immediately after the end signal should be others as alphanumeric symbols. The symbols above, those that do not correspond to this are treated as pure symbols. The blocks can be nested several times provided that they do not cross or overlap.

If when processing the additional question (tag question) the follow the following phrases at the moment when the Pointer "," indicates the structure (cluster) after "," to "?" cleared as a unit and a flag is considered a Block set. That is, the form of the set of additional questions includes:

", (Auxiliary verb) + (personal pronoun)?"
", (Auxiliary verb) n′t + (personal pronoun)?"
", (Auxiliary verb) + (personal pronoun) + not?"

Furthermore, the various auxiliary agreements include the following a: am, is, are, were, were, do, does, did, have, has, had, will, shall, would, should, can, cannot, could, may, might, must, ought, won′t, shan′t, need, dare, used. The species of the personal pronoun I, you, he, she, it, we, they a.

This is called the information in the innermost block layer, to which they belong. For example in an English sentence: you said so, didn't you ?, the whole part as a block in terms of structural composition in [You said so,] <with additional question <. Similarly, in the English sentence: I said, "You said so didn't you? ", the quoted sentence" You said so didn't you? " as a block 1 in terms of structural composition recognized, and further the whole part becomes one Block 2 recognized with regard to the structural arrangement. That is, [I said, [You said so,] <provided with an additional question <.  

The abbreviation like "didn't" is treated after it according to a predetermined table into a complete advertised form has been developed. For the Word that has a number of developed forms becomes these all spent.

In order to carry out such processing for a so-called additional question, a corresponding section 9210 is arranged between the morphological analysis section 9016 and the syntactical analysis section I 9020 in another modification of the invention. Fig. 87 shows the respective sections. In these figures, the elements which are identical to those in Fig. 64 are given the same reference numerals.

Section 9210 receives the result of the morphological analysis together with the English sentence entered from section 9016 provided for this purpose, and as shown in FIG. 88, a block is arranged for the sentence. In the examples shown in the figure, block is 0 (start: 1, end: 12, goal: sentence, role: role). In this case, in this modified embodiment, the word is represented by the number for the word. For example, in this modified embodiment, the block contains a sentence in addition to a clause and a group. In this case, the term block also includes a paragraph and the whole sentence, each of which can be considered a block.

The collective composition of the block for the input English sentence including the supplementary question may be the same as that in the flowchart described above and shown in Fig. 81. That is, processing 9300 to prepare the block of the sentence is performed before the start of processing. For example, in an English sentence I said, "It is good, isn't it?" Block 0 formed (start: beginning of the sentence, end: end of the sentence, role: sentence, goal: sentence). In the syntactic analysis section I 9020 , an analysis is performed using the same flow chart as shown in FIG. 82. Processing in the supplemental question processing section 9210 will be explained with reference to FIGS. 90A and 90B. First, a pointer to the word at the top of the word information is set ( 9370 ). If it is not a comma, the pointer is advanced step by step ( 9384 ). This is repeated until the end of the sentence ( 9371 ). It is then checked whether the word closest to the comma is a word belonging to the α group or a word belonging to the β group while the pointer is left where he is ( 9373, 9379 ). In this case, it is then determined that what belongs to an auxiliary verb, or is a verb in the language part and is not in negative form, is words belonging to the α group, while what is the negative form of the Includes auxiliary verbs or be the negative form of a verb in which the language part is words belonging to the β group. If the word does not belong to any of the groups, the pointer is advanced ( 9484 ) and the operations repeated until the end of the sentence ( 9371 ).

In the event that the word belongs to the α group, step 9384 , in which the pointer advances, is performed if the word closest to the words of the a group is not a pronoun. If it is a pronoun, it is checked whether the next word is "not" or not ( 9375 ) and if it is not "not", it is checked whether the word closest to the pronoun is a question mark or not ( 9377 ). If it is not a question mark, step 9384 is performed. If it is the question mark, the goal is written again in a "negative sentence" and the role in the "additional question sentence" for the innermost layer block ( 9378 ) and the ", .. ?" is deleted from the word information table ( 9383 ). The innermost layer block denotes blocks that meet the conditions: start position ≦ (position for ",") and which also meet the condition:
End position ≧ (position for "?") For the block position and with the minimum (end position... Start position).

If the word closest to the pronoun is "not" at step 9375 , it is checked whether the word closest to the "not" is a question mark or not ( 9376 ). If it is not a question mark, step 9384 is performed, that is, the pointer advances. If it is the question mark, the target for the innermost shift block is written in an "actuation sentence" while the role is written in an "additional question sentence" ( 9382 ), and ", .. ?" is deleted from the word information table ( 9383 ).

At step 9379 , if the word closest to the comma is a word belonging to the β group, step 9384 is performed if the word closest to the β group is not a pronoun. If it is a pronoun, a check is made to see if the word closest to it is a question mark ( 9381 ) or not, and step 9384 is entered if it is not a question mark. If it is the question mark, the target of the innermost layer block is written in a "confirmation sentence " while the role is written in a " supplementary question mark" ( 9382 ), and ", .. ?" is deleted from the word information table ( 9383 ). The pointer is then incremented ( 9384 ) and the operations repeated until the end of the sentence.

For example, Fig. 88 shows the information for the blocks and the words which are passed from the morphological analysis section 9016 to the supplementary question processing section 9210 for the English sentence: I said, "It is good, isn't it?" has been obtained. The block information for block 1 is (start: 4, end: 12, destination: optional, role: optional). If this is subjected to supplementary question processing in section 9210 , the block information for block 1 is rewritten in (start: 4, end: 12, destination: confirmation sentence, role: supplementary question sentence) and at the same time the word information which the supplementary question # 8th. . # 11 concerns, deleted.

• To Fig. 1
  1010 input section
  1018 word dictionary
  1036 Analysis rules file
  1100 input device
  1102 document file
  1104 input interface
  1106 processing section
  1108 Dictionary fetch section
  1110 Contrast elimination processing section
  1112 control section
  1116 Analysis input file
  1118 Analysis dictionary information file
  1016 Morphological analysis section
• To Fig. 2
  1010 input section
  1012 English text
  1014 Pre-Editing Section
  1016 Morphological analysis section
  1018 word dictionary
  1020 Analysis Section I
  1022 Analysis section II
  1024 structure conversion section
  1026 translation section
  1030 Reviewing Section
  1032 output section
  1034 Japanese sentence translated
  1036 Analysis rules file
  1038 control section
  1040 operation display section
• To Fig. 4
  Enter 1200 process
  1201 split
  1202 end position?
  1203 retrieve
  1204 Eliminate opposition
  1205 edition
• To Fig. 5
  1210 input
  1211 reshaping
• To FIGS. 8A to 8D
  1221 P → highest flag equal to 1?
  1223 PSAVE → end position greater than or equal to P → start position?
  1225 PSAVE → language part equal to verb?
  1226 P → language part equal to verb?
  1227 P → language part equal to verb?
  1231 P → effective input 0
  1232 PSAVE → effective input 0
  1235 end position?
• To Fig. 11
  2010 input device
  2012 Template file entered
  2014 input section
  2014 a buffer for entered string
  2016 processing section
  2016 a dictionary information preservation table
  2018 output interface
  2020 dictionary retrieval section
  2022 dictionary file
  2024 unit detection section
  2026 basic unit dictionary file
  2028 control section
• To Fig. 15
  Read 2100 sentences
  2102 Split
  2104 Is there a string?
  2106 retrieve
  2108 Is there an entrance?
  2110 Preserve dictionary information
  2112 unit detection
  2114 edition
• To Fig. 16
  Place 2200 pointer P at the top of the string
  2202 Query basic unit dictionary
  2204 Is there an entrance?
  2206 Advance to P.
  2208 Does the string exist?
  2210 Store unit information
  2212 Preserve failed information
• To Fig. 17
  3014 pre-editing section
  3100 input processing section
  3102 unit splitting section
  3104 Dictionary fetch section
  3108 Delimitation table
  3018 word dictionary
  3110, 3112, 3114, 3116, 3120, 3122, 3020 processing section
  3124 Dictionary information preservation table
• To Fig. 18
  3010 input section
  3012 English text
  3014 pre-editing section
  3016 Morphological analysis section
  3018 dictionary
  3022 analysis section II
  3024 structure conversion section
  3026 translation section
  3030 revision section
  3032 output part
  3034 Japanese sentence translated
  3036 Analysis rules file
  3038 control section
  3040 operation display section
• To FIGS. 19A and 19B
  32 Enter process
  3201 Split
  3202 end position
  3203 Search
  3209 Collective compilation
  3210 output
  3204 Is there an entrance?
  3205 Numeric flag equal to 1?
  3206 numbering
  3207 Collective compilation
  3208 conservation
  3212 with hyphen?
  3213 Process hyphenated number
  3214 currency symbol?
  3216 preservation
  3217 Delete currency symbol
  3215 Process sequence of numeric characters
• To Fig. 20
  3220 initialization process
  3221 numerical value?
  3222 Is the previous one a currency symbol?
  3223 Put together into a single noun
  3224 Is the previous one?
  3225 Put together into a single noun
  3226 advance pointer
  3227 end position
• To FIGS. 21A and 21B
  3230 Initialization process
  3231 Split
  3232 end position?
  3233 Make the stored value a numerical value
  3235 Get
  3236 Is there an input?
  3237 preservation
  3238 Numeric flag equal to 1?
  3239 numbering
  3240 addition
  3241 preservation
  3207 Collective compilation
• To FIGS. 22A and 22B
  3250 initialization
  3251 Is * P numeric?
  3252 Is * P a notation?
  3253 Is * P a decimal point?
  3254 i equals i × 10
  3255 preservation
  3256 i equal to 1?
  3257 valsave val.save * 10 + number (* P)
  3258 val.save val.save + number (* P) / i
  3259 Advance pointer P.
  3260 end position?
  3261 Make the numerical value as val - save
  3207 Collective compilation
• To FIGS. 23A and 23B
  Set 3270 pointer to previous position
  3271 Is there no word at the previous position?
  3272 input is "and"
  3273 Does the pointer display a numerical value?
  3276 end position?
  3277 Does the pointer show a numerical value?
  Round off 3278 v -now as vl
  3281 Clear "and" information
  3282 P → End position End position of the memory reference unit
  3284 preservation
• To Fig. 30
  4012 input device
  4014 word processing section
  4014 a buffer for entered string
  4016 unit splitting section
  4018 limitation table
  4020 reference dictionary
  4022 reference fetch section
  4026 Section for processing the previous sentence end
  4028 proper name processing section
  4030 Section for processing a proper name in itself
  4032 processing section
  4034 Section for obtaining missing feature information
  4036 processing section
  4036 a Dictionary information preservation table
  4038 analysis section
• To Fig. 32
  Enter 4100 process
  4102 Split
  4104 Are strings ended?
  4106 Get
  4108 Is there an entrance?
  4110 proper name?
  4112 preservation
  4114 Processing for registered proper names
  4116 capital letter?
  4118 Record the proper name as an unregistered word
  4120 Processing for an unregistered proper name
  4122 Feature addition does not comply
  4124 output
• To Fig. 33
  4200 Is the previous one an unregistered proper name?
  4202 Record whole part with feature information itself
  4204 Is the previous one a registered proper name?
  4206 Is feature information of the previous one unknown?
  4208 feature information per se is unknown
  4210 Record the whole part with the feature information from the previous one
  4212 Is there a common characteristic?
  4214 Record the proper name with the common characteristic
  4216 Record the proper name with the characteristic information itself
  4218 Record
• To Fig. 34
  4300 Is the previous one a candidate for the end?
  4302 Assess whether the previous is an end
  4303 Convert large letters to small letters
  4306 Is there an entrance?
  4308 conservation
  4310 Preservation as an unregistered proper name
  4312 Judge whether the previous is not an end
  4314 Assess whether it is a proper name in itself
  4316 Processing the proper name
• To Fig. 35
  Place 4400 hands on the upper end
  4402 proper names?
  4404 Feature information is unknown
  4406 addition of the incorrect feature information
  4408 advance pointer
  4410 end?
• To Fig. 37
  5014 pre-editing section
  51 Input processing section
  5102 unit splitting section
  5104 Dictionary fetch section
  5108 Delimitation table
  5018 reference dictionary
  5110 position information processing section
  5112 Section for judging the previous sentence end
  5114, 5116, 5118 proper name processing section
  5124 Dictionary Information Preservation Table
  5020 Morphological Analysis Section I
• To Fig. 38
  5010 input section
  5012 English text
  5014 pre-editing section
  5016 Morphological analysis section
  5018 word dictionary
  5020 Syntactic analysis section
  5022 Syntactic analysis section II
  5024 structure conversion section
  5026 translation section
  5030 revision section
  5032 output section
  5034 Japanese sentence translated
  5038 control section
  5040 operation display section
• To Fig. 40
  Enter 5200 process
  5201 Split
  5202 end position?
  5203 Get
  5204 Is there an input?
  5205 proper name?
  5206 preservation
  5207 Processing of registered proper names
  5210 output
  5212 capital letter?
  5213 Processing of unregistered names
  5214 Record the proper name as an unregistered name
• To Fig. 41
  5220 position information
  5221 pattern 0
  5222 pattern 1
  5223 pattern 2 or 3
  5230 Is the following an unregistered proper name?
  5231 POS equals 1?
  5232 POS equals 0?
  5233 Record the whole part as POS equal to 1
  5234 Record separately as POS equal to 0
  5235 Record whole part as POS equal to 0
• To Fig. 43
  5240 Is the previous one an unregistered proper name?
  5241 Convert the previous word to an unregistered one
  5242 Record separately as POS equal to 1
  5250 Is the previous one an unregistered proper name?
  5251 POS equal to 1?
  5252 POS equals 0?
  5253 POS equal self
  5254 "The"?
  5255 As POS equals POS. . . Record self with previous word
  5256 Record "3" from "the" itself as a POS
  5257 Separately as POS equals POS. . . record self
• To Fig. 45
  5260 Is the preceding one a sentence end candidate?
  5261 Judge whether the candidate is a sentence end
  5262 Convert upper case letters to lower case letters
  5263 Is there an entrance?
  5264, 5265 Record
  5221 Processing pattern 0
• To Fig. 47
  6012 input device
  6014 input processing section
  6014 a buffer for entered string
  6016 unit splitting section
  6018 delimitation table
  6020 reference dictionary
  6022 reference fetch section
  6026 Section for processing the previous sentence end
  6028 Section for processing the previous proper name
  6030 Section for processing the proper name
  6036 processing section
  6036 a Dictionary information preservation table
  6038 Syntactic analysis section
• To Fig. 49
  Enter 6100 process
  6102 Split
  6104 end?
  6106 retrieve
  6108 Is there an entrance?
  6110 proper name?
  6112 Record
  6114 Processing the registered proper name
  6116 Capital letter?
  6118 Record the proper name as an unregistered word
  6120 Processing the unregistered proper name
  6122 output for FIG. 50
  6200 Is the previous one an unregistered proper name?
  6202 Record the whole part as a proper name with characteristic information
  6204 proper name?
  6206 Is feature information unknown?
  6208 Is feature information itself unknown?
  6210 Whole part as proper names with characteristic information of the previous record
  6212 Record the proper name separately
  6214 Record
• To Fig. 51
  6300 Is the previous one a sentence end candidate?
  6302 Judge whether the previous one is an end
  Convert 6304 upper case to lower case
  6306 Is there an entrance?
  6308 Record
  6310 Record the proper name unchanged
  6312 Judge whether the previous one is not an end
  6314 Assess whether the noun itself is a proper name with unknown feature information
  6316 Processing the registered proper name
• To Fig. 53
  7014 pre-editing section
  7100 input processing section
  7102 unit splitting section
  7104 Delimitation table
  7018 dictionary
  7106 Dictionary fetch section
  7108 Morpheme processing
  7110 initial value setting section
  7112 Adjustment Retrieval Section
  7114 unit splitting section
  7116 polling section
  7118 Section for generating morpheme processing information
  7120 polling section
  7122, 7124 processing section
  7126 Dictionary information preservation table
  7020 Syntactic analysis section I
• To Fig. 54
  7010 input section
  7012 English text
  7014 pre-editing section
  7016 Morphological analysis section
  7018 dictionary
  7020 Syntactic analysis section
  7022 Syntactic analysis section II
  7040 structure forming section
  7026 translation section
  7030 revision section
  7032 output section
  7034 Japanese sentence translated
  7036 Analysis rules file
  7038 control section
  7040 operation display section
• To Fig. 55a and 55b
  7300 input
  7302 compartments
  7304 Finished?
  7306 Get
  7308 Addition of morpheme processing information
  7310 Provide information with morpheme processing?
  7312 record
  7314 collective processing
  7410 initial setting
  7412 Adjust by (p + n) th unit
  7414 adapted?
  Increase 7416 n by 1
  7418 input buffer end?
  7420 retrieval by (p + n)
  7422 addition of morpheme processing information
  7424 n <1
  7426 adjustment
  7428 adapted?
  7430 accumulation of P on (P + n -1)
  Record 7434 P separately
• To Fig. 58
  8020 input processing section
  8021 unit splitting section
  8023 limitation table
  8024 reference table
  8025 Dictionary information preservation table
  8026 dictionary retrieving section
• To Fig. 59
  8030 adjustment section
  8031 prefix dictionary
  8024 reference dictionary
  8032 Dictionary fetch section
  8033 adjustment section
  8034 Dictionary information preparation section
  8035 Derivative processing section
  8025 Dictionary information preservation table
• To Fig. 60
  8040 adjustment section
  8041 Sufix dictionary
  8042 processing section
  8043 Dictionary fetch section
  8044 adjustment section
  8045 Section for processing dictionary information preparation
  8046 dictionary retrieving section
  8047 word processing section
  8025 Dictionary information preservation table
• To Fig. 61
  8020 input processing section
  8021 unit splitting section
  8022 demarcation table
  8024 reference dictionary
  8025 Dictionary information preservation table
  8030 adjustment section
  8031 prefix dictionary
  8032 Dictionary fetch section
  8033 adjustment section
  8034 Dictionary information preparation section
  8040 adjustment section
  8041 Sufix dictionary
  8042 processing section
  8043 Dictionary fetch section
  8044 adjustment section
  8045 Section for processing dictionary information processing
  8040 adjustment section
  8047 word processing section
• To Fig. 62
  8020 input processing section
  8021 unit splitting section
  8022 demarcation table
  8024 reference dictionary
  8025 Dictionary information preservation table
  8030, 8033 adjustment section
  8031 prefix dictionary
  8032 Dictionary fetch section
  8034 Dictionary information preparation section
  8040 adjustment section
  8041 Sufix dictionary
  8042, 8046 dictionary retrieving section
  8044 adjustment section
  8045 Section for processing dictionary information preparation
  8047 word processing section
  8050 Processing section for rated information
  8051 preparation
• To Fig. 63
  8001 input section
  8002 English text
  8003 pre-editing section
  8004 Morphological analysis section
  8005 Syntactic analysis section I
  8006 Syntactic analysis section II
  8007 operation display section
  8008 word dictionary
  8009 Analysis rule file
  8010 control section
  8011 structure conversion section
  8012 Section for creating a translated sentence
  8013 revision section
  8014 output section
  8015 Japanese sentence
• To Fig. 64
  9010 input section
  9012 English text
  9014 pre-editing section
  9016 Morphological analysis section
  9018 word dictionary
  9020 analysis section I
  9022 analysis section II
  9024 structure conversion section
  9026 Section for creating a translation
  9030 revision section
  9032 output section
  9034 Translated Japanese sentence
  9036 Analysis rules file
  9038 control section
  9040 operation display section
• To Fig. 65
  9014 pre-editing section
  9016 Morphological analysis section
  9018 word dictionary
  9020 analysis section I
  9036 Analysis rule file
• To Fig. 66
  Put 9100 position at the top
  Take out 9101 word
  9102 Adjusted with upper condition?
  9103 Take out the necessary number of words
  9104 Adapted to the above condition P ?
  9105 Search
  9106 Does the end condition exist?
  9107 Register block
• To Fig. 67
  9110 Has sentence ended?
  9111 Advance position by 1
  9112 retrieve
  9113 Write word information
• To Fig. 73
  9120 Assessment of goal and role
  9121 Is there a block?
  9122 Analyze the innermost block
  9123 Select
  9124 Treat what has been selected as a single block
  9125 Analyze sentence
  9126 Select
• To Fig. 76
  9010 input section
  9012 English text
  9014 pre-editing section
  9016 Morphological analysis section
  9018 word dictionary
  9020 analysis section I
  9022 analysis section II
  9024 structure conversion section
  9026 translation section
  9030 revision section
  9032 output section
  9034 Translated Japanese sentence
  9036 Analysis rules file
  9038 control section
  9040 operation display section
  9200 processing section for information let
• To Fig. 77
  9014 pre-editing section
  9016 Morphological analysis section
  9020 analysis section I
  9018 word dictionary
  9036 Analysis rules file
  9200 processing section for information let
• To Fig. 81
  9300 Prepare a block
  Put 9100 position at the top
  9101 Process for taking out a word
  9102 Adapted to the above condition?
  9103 Take out the necessary number of words
  9104 Adapted to the above condition?
  9105 Search
  9106 Does the end condition exist?
  9107 Advertise block
• To Fig. 82
  9120 Assessment of goal and role
  9121 Is there a block?
  9122 Analyze innermost block
  9123 Select
  9124 Treat what is selected as a single block
• To FIG. 83A and 83B
  9330 Place pointer on top of word information
  9331 Is information let 0?
  9332 Is the language part a punctuation mark?
  9333 Mark innermost block
  9334 Information let
  9335 Recognize the role as a requirement record
  9336 Recognize the role as an invitation phrase
  9337 Recognize the target as a command set
  9338 Delete word information
  9339 Advance pointer by 1
  9330 end?
• To Fig. 84
  9350 end of sentence?
  9351 Advance position by 1
  9352 Get
  9353 Is there an entrance?
  9354 With hyphen?
  9355 Announcement of the block
  9356 Remove the hyphen
  9357 Get
  9358 Writing out word information
  9359 Writing out word information
• To Fig. 86
  9010 input section
  9012 English text
  9014 pre-editing section
  9016 Morphological analysis section
  9018 word dictionary
  9020 analysis section I
  9022 analysis section II
  9024 structure conversion section
  9026 translation section
  9030 revision section
  9032 output section
  9034 Translated Japanese sentence
  9036 Analysis rules file
  9038 control section
  9040 operation display section
  9210 Processing section for additional question
• To Fig. 87
  9014 pre-editing section
  9016 Morphological analysis section
  9018 word dictionary
  9020 analysis section I
  9036 Analysis rules file
  9210 Processing section for additional question
• To FIG. 90A and 90B
  Place 9370 pointer at the top
  9371 end?
  9372 comma?
  9373 group
  9374 Proper name?
  9375 "not"
  9376 Question marked?
  9377 Question marked?
  9378 Write the goal for negative sentence and the role for additional question
  9376 β group?
  9380 proper name?
  9381 Question marked?
  9382 Write the destination for the confirmation sentence and the role for the additional question
  9383 Delete "question mark"
  9384 Advance pointer by 1

Claims (30)

1. Language analyzer, characterized by a dictionary device with dictionary data stored therein, including morpheme data for words, compound words and phrases, and
analysis means for performing a morphological analysis on an input sentence with reference to the dictionary means, wherein,
the dictionary device contains data for the degree of coupling, which represents the degree of coupling between the respective words which form the compound words or phrases, and
the analyzer refers to the dictionary device with respect to the respective words contained in the input sentence and when retrieving a number of dictionary data for one word in combination with other words, the combination of words with a higher degree of coupling by referring to the data for the Select the degree of coupling. 2. Speech analyzer according to claim 1, characterized in that that if the data for the degree of coupling a high degree of coupling for a number of combinations between one retrieved in the dictionary facility Word and other words indicate the analyzer Preferred dates prepared to follow a certain rule  to give preference to one of the number of combinations. 3. Language analyzer, characterized by an input device for entering a character string of a predetermined language; a basic dictionary device used for retrieving the input string and having basic data stored therein, and
an analysis device for grammatical analysis of the input character string by retrieving the basic dictionary device, wherein
the analyzer retrieves the basic dictionary device for the input string and when thereby retrieving part of the string, retrieves the basic dictionary device for other parts in the string in the same manner to thereby morphologically analyze the string. 4. speech analyzer according to claim 3, characterized in that that the basic dictionary facility a Dictionary setup for basic units with data is to express dimensional units stored in it, and that the analysis means the dictionary means with the basic units for the entered string calls, to thereby morphologically analyze whether the String expresses a unit of measure or not. 5. Speech analyzer according to claim 4, characterized in that that the analysis device the dictionary device with basic units for the string gets, and if the string just from the combination of strings that exist in the dictionary facility stored with basic units, which is a unit as a result of retrieving expresses it as one String judged to express the unit. 6. speech analyzer according to one of claims 3 to 5,  characterized in that the analysis device has a pointer that the pointer to a character is placed at the top of the entered string, and if part of the string when retrieving the basic dictionary device for the string is retrieved which starts from the character at which the pointer has been set, the pointer is set to a string that is on the part of the retrieved string follows, and that the basic dictionary device further for the subsequent string to which the Pointer is set. 7. speech analyzer according to one of claims 3 to 6, characterized in that the entered String accessed by a common dictionary facility and not stored in the dictionary facility is. 8. Language analyzer, characterized by a dictionary device with dictionary data stored therein at respective dictionary reference units, and an analysis device to subdivide the input sentence into dictionary reference units and to carry out a morphological analysis for the dictionary reference units Dictionary facility is referred to
the dictionary means includes, as the dictionary data, a distinctive display indicating that the dictionary reference units represent numbers for the number representative dictionary reference units and
the analysis means refers to the dictionary means for the respective dictionary reference units contained in the input sentence, and
when the distinctive display is included in the retrieved dictionary data, the dictionary reference unit from which the distinctive display has been retrieved together with a dictionary reference unit which is close to the dictionary reference unit and from which the other distinguishing display is retrieved is numerical Values which are meant by the two dictionary reference units are calculated together to form a single numerical value and the dictionary reference units are formed into a single analysis unit. 9. Speech analyzer according to claim 8, characterized in that that the analysis device the analysis unit, when accompanying a dictionary reference unit, which expresses a currency symbol or a unit of dimension, together with the numerical value into one Analysis unit composed. 10. language analyzer, characterized by an input device for entering a character string of a predetermined language;
a dictionary device for retrieving the input character string from the input device;
a retriever for retrieving the dictionary device for the input string and
a feature information creating means for generating a feature information of a character string not recorded in the dictionary device and a character string whose feature information is not recorded in the dictionary device among the input character strings as a result of the retrieval by the retrieving device, the feature information creating device generates a number of feature information for the string that has no feature information. 11. A speech analyzer according to claim 10, characterized in that that the analyzer further comprises a unit subdivision device to the input String in units to get the dictionary setup split, and an analyzer to track the  String, which by the unit splitter is split after retrieving the string the retriever along with the previous string analyze. 12. Speech analyzer according to claim 11, characterized in that that if feature information is either in one string or in the previous String exists and if there is no feature information is present in the other string, the analyzer the string that has no feature information has, with the feature information of a string provided. 13. A speech analyzer according to claim 11, characterized in that that if character information in each string and is in the previous string, the analysis device has a common feature information as the feature information for this string and for returns the previous string. 14. Speech analyzer according to one of claims 2 to 4, characterized characterized that the string and the previous string of characters passed through the analyzer have been morphologically analyzed, proper names are. 15. A speech analyzer according to claim 11, characterized in that that if the string of one big written characters starts from and the previous one Follow the analysis facility at the end of the sentence the uppercase letter at the beginning of the string converted to a lower case letter, and then the dictionary device by means of the retrieval device retrieves, and if it is not in the dictionary facility as a result of the retrieval process is registered, the string analyzed as an unregistered proper name.   16. language analyzer, characterized by a dictionary device which has dictionary data stored therein for the respective dictionary reference units, and
an analyzer for dividing an input sentence into dictionary reference units and for performing a morphological analysis with reference to the dictionary unit for the dictionary reference units, wherein
the dictionary device contains, as the dictionary data, distinctive information for determining the position at which a number of proper names are permitted in a subsequent proper name set for a dictionary retrieval unit that means a proper name, and the analysis device on the dictionary device for the respective dictionary -References referenced in the input sentence, and when the distinction information is contained in the retrieved dictionary data, the dictionary reference unit from which the discrimination information was retrieved together with the dictionary reference unit which is adjacent to that dictionary reference unit and means another proper name, combined into a single analyzer according to the position determined by the discrimination information. 17. Language analyzer, characterized by an input device for entering a character string of a predetermined language;
a dictionary device for retrieving the character string input to the input device and a feature information analyzing device which retrieves the dictionary device for the input character string and then analyzes the feature information of the string, the feature information analyzing device analyzing the feature information of the string while pre-analyzing the feature information of the string and considering that string. 18. A speech analyzer according to claim 17, characterized in that that the feature information analyzer the feature information of a number of strings by arranging and assembling the number collectively Strings parsed. 19. A speech analyzer according to claim 17, characterized in that that if the feature information in this a string or in one of these strings preceding string is present and if the feature information does not exist in the other string the feature information analyzer is the string, that has no feature information with the feature information that supplies a sequence of characteristics. 20. Speech analyzer according to one of claims 17 to 19, characterized in that the string, which by the feature information analyzer has been analyzed, is a proper name. 21. Language analyzer, characterized by a dictionary device with dictionary data stored therein for each of the dictionary reference units, and
analysis means for dividing an input sentence into dictionary reference units and for performing a morphological analysis for the dictionary reference units by reference to the dictionary means, the analysis means distinguishing that a sequence of dictionary reference units with a particular semantic element is a composite Is a unit that expresses a very specific meaning, which is formed according to a certain rule and forms a sequence of dictionary reference units with the specific semantic element as a single analysis unit. 22. A speech analyzer according to claim 22, characterized in that
the dictionary device contains data to distinguish the dictionary reference unit with a very specific semantic element;
the analyzer has a corresponding distinction table to distinguish that a sequence of dictionary reference units with the specific semantic element is a composite unit which expresses a very specific meaning, which is composed according to a specific rule, and
the analyzer which retrieves dictionary means with respect to the respective dictionary reference units contained in the input sentence and forms the sequence of dictionary reference units with the specific semantic element into a single analysis unit, corresponding to the corresponding distinction table, if from the dictionary reference unit has been distinguished with the specific semantic element. 23. speech analyzer, characterized in that a grammatical feature, a semantic feature or a translated word as one not in a dictionary registered word is judged as a derivative with regard to the morphemic feature, such as a Affix is recognized. 24. Language analyzer, characterized by a first analysis device for performing a morphological analysis for an input sentence of a predetermined language;
second analysis means for performing an analysis for the sentence of the speech based on the result of the morphological analysis of the first analysis unit; a dictionary device with dictionary data of the language stored therein, which was used for the analysis by means of the first and the second analysis device, and
carry out a control device for retrieving the dictionary device and thus the corresponding analyzes of the first and second analysis devices, wherein
the first analyzing device queries the dictionary device, distinguishing the structural composition in that the feature of the input sentence of the language is judged in terms of the form thereof and judges a position that is a result of the analysis of the structural role of this compilation which is in the sentence for this compilation functions accordingly, and
the second analyzer analyzes the surface layer structure of the sentence of speech by applying a grammar rule based on the position and role evaluated and analyzing the possible interrelation of the components contained in the sentence. 25. A speech analyzer according to claim 24, characterized in that that the second analyzer is analyzing preferred for others performs when compiling in the sentence of language is included. 26. Speech analyzer according to claim 24, characterized in that that the first analyzer has the feature the sentence of language in terms of form and its Differentiates meaning and evaluates the apposition expression, which is based on the different features. 27. Language analyzer according to claim 24, characterized in that the predetermined language is English and
the dictionary device contains, as dictionary data, distinctive information distinguishing predetermined words containing "let's" and "let us", and
the first analyzer when it retrieves the dictionary to obtain the discrimination information excludes "let's" from the subject of the analysis in the second analyzer and "let us" as the subject of the analysis in the second analyzer if punctuation on the previous part is available. 28. A speech analyzer according to claim 27, characterized in that that the first analysis device the position as a command and role as an invitation to the Compilation which evaluates the excluded part contains. 29. A speech analyzer according to claim 24, characterized in that that the first analyzer if no dictionary data is included when the dictionary setup for a number of hyphenated words the dictionary facility is retrieved for each of the number Word retrieves the whole part of the number of words as one Compilation evaluated and the position of the compilation rated as an adjective group. 30. Language analyzer according to claim 24, characterized in that the predetermined language is English, and
the first analyzer distinguishes the feature of the English sentence in terms of its form, the supplementary question group distinguishes based on the differentiated feature, evaluates the whole part of the supplementary question group as a compilation whose position is a supplementary question group, and excludes it from the subject of the analysis in the second analysis device.