JP4729063B2 - Machine translation apparatus, method and program - Google Patents

Description

  The present invention relates to a machine translation apparatus, method, and program suitable for performing translation between a first language and a second language in an e-mail transmission / reception tool, a web browser, or the like.

  In recent years, the spread of personal computers has been high and the Internet environment has been improved. As a result, the exchange of information with foreign countries has become popular through bulletin boards (hereinafter referred to as web bulletin boards) using e-mail and the web (world wide web) sites. It was. Translation accuracy by machine translation devices has also been improved, and translation processing can be performed by calling a machine translation program from e-mail software or Web browser software.

  By the way, in order to improve translation accuracy, a translation method using context information may be employed. In this document translation technology, the entire document is analyzed in advance and various information is extracted, and the extracted contents are used for translation, or the sentence information translated while the document is being translated is used for subsequent sentence translation. .

In addition, in order to improve the translation of mail documents and Web bulletin boards, there is one that improves translation accuracy by extracting information that can be used for translation processing from documents having a relation other than the document to be translated. (For example, refer to Patent Document 1).
JP 2003-108553 A

  However, since the entire document and related documents are analyzed in advance and various information is extracted, and the extracted contents are used for translation. In order to determine an appropriate translation, morphological analysis, syntax / semantic analysis, from the first language There is a cost equivalent to the actual translation, such as the translation is not determined unless the semantic structure conversion step is performed in the translation processing steps of semantic structure conversion, syntax generation, and morpheme generation to the second language. End up.

  In addition, when the translated sentence information is used for the subsequent sentence translation while the document is being translated, the cost is not as high as the former method, but the translation accuracy depends on the order of appearance of the sentence. is there. For example, when considering exchange of e-mail, when replying by quoting the received mail, when replying content is written below the quoted content, when writing on the quoted content, write in the middle of the quote There are various forms such as when When translating the reply part, even if there is information that improves the translation accuracy in the quoted part, if the reply part is described on the quoted part, it will be translated without using that information. There is a problem that the translation accuracy is lower than the case described under the quotation. In addition, even when described in the middle of citation, there is a problem that the translation accuracy is lowered due to the relationship with the position where the information for improving the translation accuracy exists.

  As described above, in the exchange of electronic information such as an electronic mail or a Web bulletin board, there is a problem that even if the contents of the document are the same, the translation accuracy depends on the order of appearance of the sentences. Further, when the entire document is analyzed in advance, there is a problem that the translation processing is expensive.

  SUMMARY OF THE INVENTION An object of the present invention is to provide a machine translation apparatus that can improve the translation accuracy by using quotation marks included in exchanges such as e-mails and Web bulletin boards, and documents having the same contents are not affected by the order of appearance of sentences. It is to provide a method and a program.

A machine translation device according to the present invention includes a machine translation program, a storage device that stores a translation dictionary unit that stores vocabulary and rules necessary for translation, an input device that inputs a source text of a first language to be translated, and an output device for outputting the translated sentence of the second language after translation, the first sentence of the machine translation program Te machine translation apparatus odor and a processor for execution, the first language document entered from the entering force system The document hierarchy determination unit determines whether the number of quotes from the beginning of the sentence is a citation format hierarchy, and the document hierarchy determination unit A document storage unit for storing sentences of the first language document for each hierarchy of the citation format, and a translation for storing translation knowledge information used when translating the sentence of the first language document input from the input device Knowledge information A pay unit, said translation dictionary of vocabulary and conversion rule and translation knowledge information stored in the translation knowledge information storage unit in the hierarchy deeper order of citation form of the sentence in the first language document stored in the document storage unit The translation knowledge information used when selecting the translation word from the vocabulary and conversion rules of the translation dictionary part used for the translation and the translation rule is extracted and stored in the translation knowledge information storage part. And a translation unit for outputting to the output device .

  According to the present invention, the quotation structure included in the exchange of e-mail, Web bulletin board, etc. is used to determine the hierarchical structure of the citation format of the document, The cost required for processing can be suppressed, and in translation of a sentence with a shallow citation format, the translation accuracy can be improved without the documents having the same contents being affected by the order in which the sentences appear.

  FIG. 1 is a block configuration diagram showing a hardware configuration of a machine translation apparatus according to an embodiment of the present invention. The machine translation device 11 is realized, for example, by installing a software program such as a machine translation program in a general computer and executing the software program in the processor 13 of the arithmetic control device 12.

  The arithmetic control device 12 performs various arithmetic operations related to machine translation. The arithmetic control device 12 has a processor 13 and a memory 14. A memory translation program 15 is stored in the memory 14, and the processor 13 performs processing. When executed, the work area 16 is used. Calculation results and the like of the calculation control device 12 are displayed and output on the display device 18 that is the output device 17 and also output to the communication network via the communication control device 19.

  The input device 20 is used to input information to the arithmetic control device 12, and includes, for example, a mouse 21, a keyboard 22, a disk drive 23, and a communication control device 19. For example, the mouse 21 and the keyboard 22 are connected via the display device 18. Then, various commands are input to the arithmetic and control unit 12, and the keyboard 22, the disk drive 23, and the communication control unit 19 input a document to be translated.

  That is, the disk drive 23 inputs / outputs a file of a document to be translated to / from a storage medium, and the communication control device 19 connects the machine translation device 11 to a communication network such as the Internet or a LAN. The communication control device 19 is a device such as a LAN card or a modem, and data transmitted / received to / from the communication network via the communication control device 19 is transmitted / received to / from the arithmetic control device 12 as an input signal or an output signal. Further, a hard disk drive (HDD) 24 is provided for storing a calculation dictionary of the calculation control device 12 and a translation dictionary storing knowledge and rules necessary for translation.

  FIG. 2 is a functional block diagram of the machine translation apparatus 11 according to the embodiment of the present invention. Each functional block in the arithmetic and control unit 12 shown in FIG. 2 corresponds to each program constituting the machine translation program 15 described above. That is, when the processor 13 executes each program constituting the machine translation program 15, the arithmetic control device 12 functions as each functional block. Each block of the storage device 25 corresponds to a storage area of the memory 14 and the hard disk drive 24 in the arithmetic control device 12.

  The machine translation device 11 includes a control unit 26 that controls the entire apparatus, an input unit 27 that interfaces with an external input, an output unit 28 that interfaces with an external output, and an input obtained via the input unit 27. It is determined whether or not a quotation mark exists at the head of the sentence of the first language document input from the translation section 29 and the input section 27 for translating the document. Document hierarchy determination unit 30 that determines the number of symbols as a citation format hierarchy, translation dictionary unit 31 that stores various translation knowledge used by translation unit 29 at the time of translation, and each citation format layer determined by document hierarchy determination unit 30 Document storage unit 32 for storing a sentence of the first language document, translation knowledge information storage unit 33 for storing translation knowledge information useful for selecting a translation word when translating a sentence related to a sentence in which a quotation mark exists, citation Record It is constituted by a reference symbol storage unit 39 for storing.

  The input unit 27 receives the first language document or command to be translated through the communication control device 19 such as the Internet as the input device 20, the keyboard 22, or the like. The control unit 26 sends the sentence data of the first language document to be translated sent from the input unit 27 to the document hierarchy determination unit 30 or sends the document extracted from the document storage unit 32 to the translation unit 29. It is responsible for overall control, such as sending the translation result of to the output unit 28.

  The translation unit 29 performs translation processing on the text data in the first language document sent from the control unit 26 while referring to the dictionary of the translation dictionary unit 31 and the knowledge information stored in the translation knowledge information storage unit 33. Is what you do.

  The translation dictionary unit 31 stores translation knowledge necessary for the translation process of the translation unit 29. For example, if the translation unit 29 performs bi-directional translation processing between English and English, the dictionary unit 31 has a basic dictionary unit 34, a technical term dictionary unit 35, and a user dictionary unit 36 for each translation direction. is doing. The basic dictionary unit 34 includes a vocabulary unit 34a, a morphological analysis rule 34b, a syntax / semantic analysis rule 34c, a conversion rule 34d, a syntax generation rule 34e, and a morpheme generation rule 34f. The technical term dictionary unit 35 includes only the vocabulary unit 35a. The user dictionary unit 36 includes a vocabulary unit 36a and a translated word learning unit 36b. The basic dictionary unit 34, technical term dictionary unit 35, and user dictionary unit 36 are configured as follows.

(1) Basic dictionary part (a) Vocabulary part Utilization information, semantic information, field information, translation information, field information for each translation word at least for each vocabulary in the first language (b) Morphological analysis rules Input sentence in the first language (C) Syntax / semantic analysis rules Knowledge for syntactic / semantic analysis after morphological analysis of input sentences in the first language (d) Conversion rules Syntax / semantic analysis Knowledge for converting the resulting semantic structure of the first language into the semantic structure of the second language (e) Syntax generation rules Knowledge for generating a word string of the second language from the semantic structure of the second language (f) Morpheme Generation rule Knowledge to output the final translation reflecting the use of words in the second language (2) Technical term dictionary part The technical term dictionary part is provided with a plurality of field-specific dictionaries. Select a dictionary to use for translation according to the contents Door can be. Consists only of vocabulary.

(3) User dictionary part The user dictionary part is a user-defined dictionary. The user can create a plurality of documents according to the document to be translated. There is a vocabulary part and a translation learning part for each.

(A) Vocabulary section Utilization information, semantic information, translated word information, etc. for terms newly registered by the user (b) Translated word learning section Stores information on translated words learned by the user for a certain first language phrase.

  Next, when referring to the translation dictionary unit 31 in the translation unit 29, the translation knowledge information storage unit 33 stores the field information when specific field information is given to the headword that is referred to. Information that is determined to be applicable to other sentences in the course of translation, such as a pair of headwords and translated words determined by the co-occurrence relationship, is stored. For example, translation knowledge information useful for selecting a translation word when translating another sentence related to a sentence having a quotation mark is stored. Details will be described later using a specific example.

  The output unit 28 outputs the second language sentence (translated sentence) translated by the translation unit 104 in accordance with an instruction from the control unit 26. The document hierarchy determination unit 30 determines whether or not the citation format hierarchy information is included in the input data passed from the control unit 26, and stores the citation format hierarchy in the document storage unit 32. The document storage unit 32 stores the sentence of the first language document for each layer of the citation format determined by the document layer determination unit 30. When the processing in the document hierarchy determination unit 30 is completed, the storage content of the document storage unit 32 is taken out from the control unit 26 and transferred to the translation unit 29 for translation. Details of these processing contents will be described later.

  FIG. 3 is a flowchart showing an example of the processing contents from the input of the first language document of the machine translation device according to the embodiment of the present invention to the output of the translation. In the following description, a case where the first language document input from the input device 20 via the input unit 27 is an original sentence in English and is translated into a Japanese translation of the second language will be described.

  When the original text data of the first language document is input via the input unit 27, the control unit 26 activates the document hierarchy determination unit 30, and the document analysis determination unit 30 determines the hierarchy of the document citation format (S101). ). Here, it is assumed that the following original text is input from the input unit 27.

He is a reliever.
> We got Taro Toshiba.
> Do you know him?
The ">" at the beginning of the second and third sentences is an email quoting symbol, and the reply contents are described in the first sentence relative to the contents of the second and third sentence mails. . Here, although the quote symbol is “>”, the quote symbol may differ depending on the mail tool. Therefore, the quote symbol may be freely set in the quote symbol storage unit 39 by the user of the machine translation apparatus. The designation of the quotation mark is not limited to a single one, and a plurality of combinations may be designated.

  In the embodiment of the present invention, as shown in FIG. 4, the quote symbol storage unit 39 is designated with “>” and “|” as quote symbols in advance, and the arrangement position M of the quote symbols is M = When 1, the quotation mark is “>”, and when the arrangement position M of the quotation mark is M = 2, the quotation mark is “|”.

  FIG. 5 is a flowchart showing the processing contents of the document hierarchy determination unit 30 in step S101 in FIG. When the original text data of the first language document is input, the document hierarchy determination unit 30 acquires the sentence S from the input character string for each line break (S201). Subsequently, a variable L indicating the hierarchy of the document citation format, a variable N indicating the determination start position of the acquired sentence S, and a variable M indicating the arrangement position of the quotation marks are respectively set to an initial value 1 (S21). When the variable L indicating the hierarchy of the citation format is the initial value (L = 1), the sentence S without the quotation mark, and the variable N indicating the determination start position of the acquired sentence S is the initial value (N = 1) Is a case where the quote symbol is “>” when the variable M indicating the start position of the sentence S and the quote symbol array position is the initial value (M = 1).

  Next, the quote symbol “>” indicated by the variable M (M = 1) at the quote symbol array position is compared with the character at the head position (N = 1) of the sentence S to determine whether or not they match (S203). ). If they match, L = L + 1 is performed, and the variable L indicating the hierarchy of the citation format is deepened by one (S204). Then, N = N + 1 is performed, the determination start position of the sentence S is shifted to the next character position (second character position), and is matched with the quote symbol “>” of the quote symbol arrangement position M (M = 1)? It is determined whether or not (S205). If they match, L = L + 1 is performed, the variable L indicating the hierarchy of the citation format is deepened by one (S206), and the process returns to step S205. The processes in S205 and S206 are repeated until it is determined that the character at the determination start position (Nth position) of the sentence S does not match the quotation mark “>” at the quotation mark arrangement position M (M = 1). .

  If it is determined in step S205 that the character at the determination start position (Nth position) of the sentence S does not match the quotation mark “>” at the quotation mark arrangement position M (M = 1), then Is stored in the document storage unit 32 as a sentence of the hierarchy L (S207). Then, it is determined whether or not it is the last sentence S of the original text data of the first language document. If it is not the last sentence S, the process returns to step S201. The unit 30 returns the processing to the control unit 26.

  On the other hand, if it is determined in step S203 that the quote symbol “>” indicated by the variable M (M = 1) of the quote symbol array position does not match the character at the head position (N = 1) of the sentence S, M = M + 1 is performed, another quotation symbol “|” is obtained from the arrangement position M (M = 2) of the quotation marks, and compared with the character at the head position (N = 1) of the sentence S (S209). If they do not match, the variable L indicating the hierarchy of the citation format at that time is stored in the document storage unit 32 as a sentence of the hierarchy L (S207).

  On the other hand, if they match, L = L + 1 is performed, and the variable L indicating the hierarchy of the citation format is deepened by one (S210). Then, N = N + 1 is performed, the determination start position of the sentence S is shifted to the next character position (second character position), and is matched with the quote symbol “|” at the quote symbol arrangement position M (M = 2). It is determined whether or not (S211). If they match, L = L + 1 is performed, the variable L indicating the hierarchy of the citation format is deepened by one (S212), and the process returns to step S211. When it is determined that the character at the determination start position (Nth position) of the sentence S does not match the quotation mark “|” at the quotation mark arrangement position M (M = 2), The variable L indicating the hierarchy is stored in the document storage unit 32 as a sentence of the hierarchy L (S207). In this way, it is determined from the head of the sentence S whether or not it matches each quote symbol stored in the quote symbol array.

  For example, since there is no quotation mark in the first sentence “He is a reliever.”, The process proceeds to steps S201, S202, S203, S209, and S207. In step S207, the sentence S is stored in the document storage unit 32 together with the variable L (L = 1) indicating the hierarchy. Subsequently, in step S208, it is determined whether or not the last sentence has been read. If there is still a document, the process returns to step S201 to acquire the next sentence S. Then, each variable L, N, M is initialized in the same manner as described above.

  In the case of “We got Taro Toshiba” in the second sentence, in the determination in step S203, “>” in the quote symbol array matches “>” as the character indicated by N in the sentence S, and the process proceeds to S204. . In S204, L = L + 1 is performed, and the variable L indicating the citation format hierarchy is deepened by one. Then, N = N + 1 is performed, the determination start position of the sentence S is shifted to the next character position (second character position), and is matched with the quote symbol “>” of the quote symbol arrangement position M (M = 1)? It is determined whether or not (S205). When there is no part that matches the quotation mark, the process proceeds to S207, and the sentence S is stored in the document storage unit 32 together with the variable L indicating the hierarchy. The same process is performed for the third sentence.

  FIG. 6 is an explanatory diagram of an example of a first language document stored in the document storage unit 32. As shown in FIG. 6, the sentence of the first language document to be translated is stored for each hierarchy of the citation format. That is, “He is a reliever.” Is stored in level 1 for the first sentence without a quotation mark, the second sentence “We got Taro Toshiba.” With one quotation mark, and “Do you know him?” "Is stored in level 2. In this case, the hierarchy 1 is shallower than the hierarchy 2, and conversely, the hierarchy 2 is deeper than the hierarchy 1. The document storage unit 32 also stores a translation of the second language translated by the translation unit 29 in a later process.

  In the above description, the hierarchy of the document citation format is determined based on the citation symbol, but in the case of forwarded mail, “- ------ Original Message -------- "" and "<zzzz@bbb.co.jp> wrote:" may be included. You may utilize these as judgment of a hierarchy.

  When the process of step S101 in FIG. 3 is completed, the control unit 26 next acquires the maximum value Lmax of the hierarchy from the document storage unit 32 (S102). Subsequently, a sentence corresponding to the maximum value Lmax of the hierarchy is acquired from the document storage unit 32 (S103). In this case, if there are multiple parts of the same hierarchy, all sentences are acquired. The translation unit 29 is activated and translated for the acquired sentence data (S104). In other words, translation is performed from the deepest level. In the example of the document storage unit 32 illustrated in FIG. 6, the maximum value of the citation format layer L is 2. And the sentence of the maximum value (L = 2) of the hierarchy L includes two sentences of the second sentence “We got Taro Toshiba.” And the third sentence “Do you know him?”. When there are a plurality of sentences in the maximum value of the hierarchy, the sentences are to be translated in the order in which the sentences of the first language document appear. Therefore, in this case, “We got Taro Toshiba.” Is the first sentence to be translated.

  FIG. 7 is a flowchart showing the translation processing of the translation unit 29 in step S104 in FIG. FIG. 7 shows the translation processing of the translation unit 29 for one sentence in the original text of the first language document. For the translation of the entire input original text, steps S301 to S308 in FIG. 7 are repeated until the end of the document. The

  When the original text to be translated is sent from the control unit 26, the translation unit 29 first reads the translation knowledge information from the translation knowledge information storage unit 33 (S301). FIG. 8 is an explanatory diagram of an example of translation knowledge information stored in the translation knowledge information storage unit 33. The translation knowledge information includes field information 37 that identifies the field of the document to be translated, headwords, parts of speech, translation selection information 38 of the translations determined by application of conversion rules, and the like. In the initial state, both are empty.

  The translation unit 29 performs dictionary lookup and morphological analysis processing on the second input sentence “We got Taro Toshiba.” Using the following knowledge of the dictionary unit 31 (S302).

(1) Basic dictionary 34
(A) Vocabulary part 34a
(B) Morphological analysis rule 34b
(2) Technical term dictionary part 35
(3) User dictionary unit 36
Next, in step S303 of FIG. 7, the translation unit 29 performs syntax / semantic analysis of the input original sentence using the syntax / semantic analysis rule 34c of the basic dictionary unit 34. At this stage, the structure of the input language analysis result is established.

  Next, the translation unit 29 shifts the processing to step S304 in FIG. 7, and the vocabulary unit 34a of the basic dictionary unit 34, the conversion rule 34d, the vocabulary unit 35a of the technical term dictionary unit 35, and the vocabulary unit of the user dictionary unit 36. 36a and the translation learning unit 36b are used to convert the structure of the input language into the structure of the output language. At this stage, the translation of the input word is determined together with the linguistic structure of the output language.

  FIG. 9 is an explanatory diagram of an example of the vocabulary part 34a of the basic dictionary part 34 for the input character string “Taro Toshiba.” Of the second input sentence “We got Taro Toshiba.”. The translation unit 29 refers to the entry word field information 37a from the vocabulary part 34a of the basic dictionary part 34 for the input character string “Taro Toshiba.”, And determines the presence or absence of the entry word field information 37a (S305). If the headword field information 37a exists, the headword field information 37a (sports: baseball) is stored in the translation knowledge information storage unit 33 (S306). As a result, as shown in FIG. 8, “sports: baseball” given to “Taro Toshiba” of the input sentence is stored in the translation knowledge information storage unit 33.

  In the example sentence according to this example, the field information 37 is stored in the translation knowledge information storage unit 33. However, not only the field information 37 but also a translation word is determined based on the relationship (translation word selection information) with a word used together. You may make it store as translation knowledge information. This will be explained using the following example sentence.

Example sentence `` He erected a tent city along the banks of the Grand Canal. ''
In this example sentence, “bank” includes translations of completely different properties, “bank” and “bank”. In such a case, the following rules 1, 2, and 3 are described in the conversion rule 34d of the basic dictionary unit 34, for example.

Rule 1 ("Bank" if the object of "of" is a city name)
Rule 2 (If the object of “of” is “* canal” or “* river”, “bank”)
Rule 3 (“Bank” if none of these apply)
In this example sentence, “Grand Canal” exists, so rule 2 is adopted, and the translation is determined to be “bank”. Information determined in this way is also stored in the translation knowledge information storage unit 33. When the structure of the output language is constructed in this way, the translation unit 29 moves the process to step S307, and uses the syntax generation rule 34e of the basic dictionary unit 34 to convert the two-dimensional structure into a one-dimensional structure. Convert. Finally, in step S308, the translation unit 29 uses the morpheme generation rules 34f of the basic dictionary unit 34 to generate the surface layer form of each word, and outputs the final translation to the document storage unit 32.

  Subsequently, the third sentence “Do you know him?”, Which is the next sentence, is also processed in steps S301 to S308, and the translated sentence is output to the document storage unit 32. When the translation is output to the document storage unit 32, the translation of the hierarchy 2 is completed and the process returns to the control unit 26.

  Next, in step S105 in FIG. 3, the control unit 26 determines whether the translation of the citation format layer 1 has been completed (S105). Since the layer (L) in the quotation form “Do you know him?” That has been translated is not layer 1 (layer 2), L = L−1 is executed (S106), and the process proceeds to step S103. To migrate. As a result, the first sentence “He is a reliever.”, Which is a sentence in layer 1, is acquired, and the translation unit 29 performs the steps in FIG. 7 on “He is a reliever.”, Which is a sentence in layer 1. The translation processing of S301 to S308 is performed.

  That is, the translation unit 29 acquires from the translation knowledge information storage unit 33 the translation knowledge information that the field is “sports: baseball” from the field information 37 of FIG. 8 stored when the document of the hierarchy 2 is translated (S301). ). Subsequently, using the knowledge of the dictionary unit 31 for the input sentence “He is a reliever.”, Dictionary lookup and morpheme analysis processing (S302), syntax and semantic analysis are performed (S303), and the conversion processing of step S304 is performed. Transition. When the translation is determined in step S304, it is determined whether there is a translation in the same field with reference to the acquired translation knowledge information “Sports: Baseball” and the translation field information 37b in FIG.

  Here, when there are a large number of translated words in the headword, the headword field information 37a cannot be uniquely determined. Therefore, “0” is described in the headword field information 37a for such headword, Translation field information 37b is described for each translation word.

  Thus, when the translated word “reliever” is determined, “relief pitcher” whose translated word field information 37b is “sports: baseball” is acquired. Subsequently, it is determined whether there is newly acquired translation knowledge information (S305). If there is translation knowledge information, translation knowledge is stored in the translation knowledge information storage unit 33 (S306), and syntax generation processing is performed (S307). ), Morpheme generation processing is performed (S308), and the translation is stored in the document storage unit 32.

  When the translation of “He is a reliever.”, Which is a sentence in layer 1, is completed, the control unit 26 obtains the translation result from the document storage unit 32 (S107) as shown in step S107 in FIG. Call 28. The output unit 28 outputs the passed translation result (S108).

  FIG. 10 is a flowchart showing another example of the processing contents from when the first language document of the machine translation apparatus according to the embodiment of the present invention is input until the translation is output. In another example, in contrast to the example shown in FIG. 3, the document hierarchy determination unit 30 performs the sentence of the first language document input from the input device 20 via the input unit 27 in the document hierarchy determination processing in step S401. In the document storage unit 32, the control unit 26 sorts the translated sentences in the same order as the appearance order of the sentences constituting the document in the first language in the translation sorting process in step S402, and outputs the output unit 28. In the translation result output process in step S403, the translation result sentences are output in the same order as the appearance order of the sentences constituting the first language document. The same processes as those in FIG. 3 are denoted by the same reference numerals, and redundant description is omitted.

  When the original text data is input via the input unit 27, the control unit 26 activates the document hierarchy determination unit 30 (S401). Here, it is assumed that the following original text is input from the input unit 27.

No, He is reliever.
>> We got Taro Toshiba.
>> He is a pitcher.
> Is he a starter?
FIG. 11 is a flowchart showing the processing contents of the document hierarchy determination unit 30 in step S401 in FIG. FIG. 11 provides a variable P for storing the input original text line number in step S501, and updates the input original text line number variable P in step S502, for the example shown in FIG. In step S507, the sentence is stored in the document storage unit 32 together with the line number P as a sentence in the citation format layer L. The same processes as those in FIG. 5 are denoted by the same reference numerals, and redundant description is omitted.

  In FIG. 11, first, a variable P that stores the line number of the input original text is set to 0 (S501). Subsequently, the sentence S is acquired (S201), then the variable L indicating the document hierarchy, the variable N indicating the determination start position of the acquired sentence S, and the variable M indicating the position of the quote symbol array are set to 1, respectively. At the same time, 1 is added to P (S502).

  Then, in steps S203 to S206 and steps S209 to S212, the citation format hierarchy L is obtained by the same processing as in FIG. 3, and the line number P is stored in the document storage unit 32 together with the sentence of the hierarchy L in step S507. .

  FIG. 12 is an explanatory diagram of an example of the first language document stored in the document storage unit 32 in another example of the machine translation device according to the embodiment of the present invention. As shown in FIG. 12, the document storage unit 32 stores the citation format hierarchy, the original text, and the line number where the original text existed. It is determined whether or not processing has been completed up to the last sentence (S208), and if processing has been completed up to the last sentence, the process returns to the control unit 26.

  In this way, the process of steps S102 to S107 of FIG. 10 is performed in the same manner as in FIG. Then, the control unit 26 acquires line numbers together with the translation from the document storage unit 32, sorts the acquired data in ascending order based on the line numbers (S402), calls the output unit 28, and outputs the output unit 28. Outputs the passed translation results in the same order as the appearance order of each sentence constituting the document in the first language (S403).

  According to the embodiment of the present invention, a translation obtained by judging the hierarchical structure of a document using quotation marks included when exchanging e-mails, Web bulletin boards, etc., and translating in descending order of hierarchy. Knowledge information can be applied to translation of sentences with a shallow hierarchy. Accordingly, it is possible to improve translation accuracy without affecting documents having the same content in the order of appearance of sentences.

  In addition, since translation is performed in the order of deeper hierarchy, depending on the description state of the original sentence, if the output is performed in the translated order, the original sentence and the translated sentence may be arranged differently. Therefore, by storing the order of appearance of the original sentences and sorting the corresponding translated sentences in the order of the original sentences, the translated sentences can be output in the same order as the original sentences, so that the translation results are easy to read.

The block block diagram which shows the hardware constitutions of the machine translation apparatus which concerns on embodiment of this invention. The functional block diagram of the machine translation apparatus concerning embodiment of this invention. The flowchart which shows an example of the processing content after inputting the 1st language document of the machine translation apparatus which concerns on embodiment of this invention until it outputs a translation. Explanatory drawing of the arrangement position of the reference symbol in embodiment of this invention. FIG. 4 is a flowchart illustrating processing contents of a document hierarchy determination unit in step S101 in FIG. Explanatory drawing of an example of the 1st language document stored in the document memory | storage part in embodiment of this invention. The flowchart which shows the translation process of the translation part in step S104 in FIG. Explanatory drawing of an example of the translation knowledge information stored in the translation knowledge information storage part in embodiment of this invention. Explanatory drawing of an example of the vocabulary part of the basic dictionary part with respect to input word "Taro Toshiba." Of input sentence "We got Taro Toshiba.". The flowchart which shows another example of the processing content after inputting the 1st language document of the machine translation apparatus which concerns on embodiment of this invention until it outputs a translation. 11 is a flowchart showing processing contents of a document hierarchy determination unit in step S401 in FIG. Explanatory drawing of an example of the 1st language document stored in the document memory | storage part in another example of the machine translation apparatus which concerns on embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 ... Machine translation apparatus, 12 ... Operation control apparatus, 13 ... Processor, 14 ... Memory, 15 ... Machine translation program, 16 ... Work area, 17 ... Output device, 18 ... Display apparatus, 19 ... Communication control apparatus, 20 ... Input Device 21 ... Mouse 22 ... Keyboard 23 ... Disk Drive 24 ... Hard Disk Drive 25 ... Storage Device 26 ... Control Unit 27 ... Input Unit 28 ... Output Unit 29 ... Translation Unit 30 ... Document Hierarchy Determination , 31 ... Translation dictionary part, 32 ... Document storage part, 33 ... Translation knowledge information storage part, 34 ... Basic dictionary part, 35 ... Technical term dictionary part, 36 ... User dictionary part, 37 ... Field information, 38 ... Translation word selection Information, 39 ... Quote symbol storage

Claims (4)

A machine translation program, a storage device storing a translation dictionary part storing vocabulary and rules necessary for translation, an input device for inputting the original text of the first language to be translated, and a translated text of the second language after translation an output device for outputting, the machine translation program Te machine translation apparatus odor and a processor for execution, whether the top reference symbol sentence of the first language document entered from the entering force device is present When there is a citation symbol, the document hierarchy determination unit determines the number of citation symbols from the beginning of the sentence as the citation format hierarchy, and the first for each citation format hierarchy determined by the document hierarchy determination unit. A document storage unit for storing a sentence of a language document, a translation knowledge information storage unit for storing translation knowledge information used in translating a sentence of a first language document input from the input device, and the document storage Part Hierarchy citation form of the sentence in the first language document deep forward to the translation dictionary of words and conversion rules and the translation knowledge information storing unit by using the translation knowledge information stored in performs translation the translation A translation unit that extracts translation knowledge information used when selecting a translation word from the vocabulary and conversion rules of the translation dictionary unit used at the time, stores the translation knowledge information in the translation knowledge information storage unit, and outputs a translation result to the output device. A machine translation device characterized by that.   The document hierarchy determination unit stores the order of appearance of sentences of the first language document input from the input device in the document storage unit, and the output unit includes the order of appearance of sentences constituting the document of the first language. The machine translation apparatus according to claim 1, wherein sentences of translation results are output in the same order. A machine translation program, a storage device storing a translation dictionary part storing vocabulary and rules necessary for translation, an input device for inputting the original text of the first language to be translated, and a translated text of the second language after translation An output device for outputting and a processor for calculating and executing the machine translation program, and the storage device is configured to calculate the number of citation symbols present at the beginning of a sentence of the first language document input from the input device in a citation format. A document storage unit for storing sentences of the first language document for each hierarchy of the citation format as a hierarchy, and translation knowledge information used when translating sentences of the first language document input from the input device In a machine translation method for performing machine translation using a machine translation device having a translation knowledge information storage unit, a sentence of a first language document input from the input device is input to the processor, The processor determines whether or not there is a quotation mark at the beginning of the sentence of the first language document input from the input device, and when there is a quotation mark, the number of quotation marks from the beginning of the sentence And the sentence of the first language document is stored in the document storage unit for each hierarchy of the determined citation format, and the citation format hierarchy is included in the sentences of the first language document stored in the document storage unit. from a deep order of the dictionary portion vocabulary and conversion rule and using a translation knowledge information stored in the translation knowledge information storage unit performs translation, lexical and conversion rules of the translation dictionary unit used in the translation A machine translation method comprising: extracting translation knowledge information used when selecting a translation word, storing the translation knowledge information in the translation knowledge information storage unit, and outputting the translated translation result to the output device. A machine translation program, a storage device storing a translation dictionary part storing vocabulary and rules necessary for translation, an input device for inputting the original text of the first language to be translated, and a translated text of the second language after translation In a machine translation program used in a computer comprising an output device for outputting and a processor for executing and executing the machine translation program, the storage device stores a quotation mark present at the head of a sentence of the input first language document. Document storage unit for storing sentences of the first language document for each citation format hierarchy, with the number as the citation format hierarchy, and translation knowledge used when translating the sentence of the first language document input from the input device A translation knowledge information storage unit for storing information; a procedure for inputting a sentence of a first language document input from the input device to the computer; and the input device The procedure for determining whether or not there is a citation symbol at the beginning of the sentence of the first language document input from and determining the number of citation symbols from the beginning of the sentence as the citation format hierarchy, , A procedure for storing the sentence of the first language document in the document storage unit for each hierarchy of the determined citation format, and in the descending order of the citation format hierarchy among the sentences of the first language document stored in the document storage unit From the vocabulary and conversion rules of the translation dictionary unit and the procedure for performing translation using the translation knowledge information stored in the translation knowledge information storage unit, and from the vocabulary and conversion rules of the translation dictionary unit used for the translation A machine translation program for executing a procedure of extracting translation knowledge information used when selecting a translation word and storing the translation knowledge information in the translation knowledge information storage unit and a procedure of outputting a translated translation result to the output device.

Images