JPH1166068A - Machine translation device/method and recording medium with recorded machine translation program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To correctly translate abbreviation symbols in parentheses when they exist in an original by analyzing the constitution of the original based on the result of a parenthesis part recognition means recognizing a word part as a noun phrase and a division means and generating the translated sentence of target language based on an analyzed result. SOLUTION: A translation control part 14 takes out one sentence being a translation object from an original/translated sentence storage part 2 by the instruction of translation execution and sends it to a morpheme analysis part 3. A dictionary retrieval part 4 retrieves the registered word of a word dictionary 12 for a morpheme which the morpheme analysis part 3 analyzes and obtains necessary data. When a character string in parentheses is the head characters of several words existing before the parenthesis writing, a parenthesis recognition part 5 treats it as a noun phrase. When plural parts of speech are considered, a multiple parts of speech dissolution part 6 decides the rule of analysis by a rule dictionary 13. A syntax analysis part 7 specifies syntax in accordance with the rule and a conversion part 9 converts the objective sentence of translation into the system of target language. A target language generation part 10 generates the translated sentence.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a machine translation apparatus and a machine translation method for translating one language into another, such as a translation from English to Japanese, and a recording medium on which a machine translation program is recorded. .

[0002]

2. Description of the Related Art In recent years, with the spread of personal computers and the development of information and communication networks, internationalization of information has been rapidly progressing, and information in other languages has become very popular. Opportunities to create documents in other languages to transmit information on their own are also increasing.
In order to utilize and transmit such information, opportunities for using a machine translation device are increasing.

[0003] In a conventional machine translator, in order to enhance the analysis function when an input sentence includes a parenthesis symbol, Japanese Patent Laid-Open Publication No. Hei 4-283864 discloses that a sentence with parentheses is divided into parentheses inside and outside parentheses. It is disclosed that the translation process is performed. Also, Japanese Patent Application Laid-Open No. 7-200588 discloses analyzing the role of parentheses outside the parentheses.

[0004]

The information that flows on the Internet and the like often relates to companies and products, and many proper nouns appear. In addition, in order to eliminate the redundancy of sentences, the abbreviations are written together, and sentences are often written using the abbreviations.

However, in a conventional machine translator, when a symbol indicating an abbreviation is given in parentheses in a sentence, the information cannot be used, and an incorrect translation result may be output. Was.

For example, as shown in FIG.
Consider an English-Japanese machine translator that translates an English sentence "Protocol Council (SPC) Information" into a Japanese sentence.

First, a morphological analysis of the original sentence is performed, and the original sentence is divided into words, and an analysis result as shown in FIG. 27 is obtained.

Next, each word subjected to morphological analysis is subjected to dictionary lookup, and a dictionary lookup result as shown in FIG. 28 is obtained. As a result, an analysis state as shown in FIG. 29 is obtained,
Based on this, multi-part-of-speech cancellation for specifying a plurality of parts of speech of a word is performed. The word "protocol" has two possibilities, a verb (plan) and a noun (protocol), but considering the structure of the entire sentence, S (Security) + V (Protocol)
+ O (Council (SPC) Information)
Most likely a verb. Therefore, an analysis result as shown in FIG. 30 is obtained.

When translation is advanced based on the analysis result, FIG.
1 is obtained. Therefore,
As shown in FIG. 32, an erroneous translation result such as "Safety formulates council (SPC) information" is output.

The present invention provides a machine translation apparatus, a machine translation method, and a recording medium on which a machine translation program can output a correct translation result when a symbol indicating an abbreviation is given in parentheses in the original sentence. The purpose is to provide.

[0011]

SUMMARY OF THE INVENTION In order to solve this problem, the present invention provides an original text storing means for storing an input original text, a dividing means for dividing the original text stored in the original text storing means into morphemes, A dictionary search unit for searching for a part of speech or a translation of a morpheme divided by the division unit with reference to the dictionary;
When the character string in parentheses is the first character of some words existing before the parentheses, a parenthesis recognition unit that recognizes the word portion as a noun phrase, a division unit, a dictionary search unit, Based on the result of the parenthesis recognition means, referring to a rule dictionary to analyze the syntax of the original sentence; a target language generating means for generating a translated sentence of the target language based on the analysis result of the original text analysis means; Output means for outputting the translated sentence generated by the language generation means.

Thus, a machine translation device capable of outputting a correct translation result when a parenthesis symbol indicating an abbreviation is given in the original sentence is obtained.

[0013]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention according to claim 1 of the present invention provides an original text storing means for storing an input original text, a dividing means for dividing the original text stored in the original text storing means into morphemes, and a word dictionary. , The dictionary search means for searching for the part of speech or translation of the morpheme divided by the division means, and the character string in parentheses is the first character of some words existing before the parentheses When the parenthesis recognition means to recognize the word portion as a noun phrase, based on the results of the dividing means, dictionary search means and parenthesis recognition means, by referring to the rule dictionary, the original sentence analysis means to analyze the syntax of the original sentence And a target language generating means for generating a translation of the target language based on the analysis result of the source text analyzing means, and an output means for outputting a translation generated by the target language generating means. To If the symbol indicating the abbreviations in arc writing is given, handling group of words that abbreviations corresponding as a noun phrase.

According to a second aspect of the present invention, in the original sentence storage means for storing the input original sentence, and in the entire original sentence stored in the original sentence storage means, the character string in parentheses is replaced by the parenthesized character string. Is also the first character of some preceding words, a bracket part registration means for registering the word part as a noun phrase in a bracket table, and a division for dividing the original sentence stored in the original sentence storage means into morphemes Means, a dictionary search means for searching for a part of speech or a translation of a morpheme divided by the division means with reference to the word dictionary, and a parenthesis registration means,
An original sentence for analyzing the syntax of the original sentence by referring to the rule dictionary based on the results of the parentheses acknowledgment means for acknowledging a word registered in the bracketed table as a noun phrase, and the results of the division means, dictionary search means and parenthesis qualification means Analysis means, a target language generation means for generating a translation of the target language based on the analysis result of the original text analysis means, and an output means for outputting the translation generated by the target language generation means, by having a configuration having Even if there is no part corresponding to the abbreviation in parentheses in one sentence to be translated, if there is a corresponding part in the text, the corresponding part is treated as a noun phrase.

According to a third aspect of the present invention, in the first or second aspect, the target language generating means generates a translated sentence of the target language based on the result of the analysis by the original text analyzing means. In this case, the translation corresponding to the abbreviation in the parentheses is represented in katakana by generating the translation corresponding to the parentheses in katakana.

[0016]

Embodiments of the present invention will be described below with reference to the drawings.

(Embodiment 1) FIG. 1 is a functional block diagram of a machine translation apparatus according to a first embodiment of the present invention, and shows the machine translation apparatus of this embodiment by functional means.

In this embodiment, the source language is English,
Although an English-Japanese machine translation apparatus that performs machine translation using Japanese as the target language has been described as an example, the source language and target language are not limited to these, and other languages may be used.

In FIG. 1, reference numeral 1 denotes an input means for inputting an original sentence and a translation instruction, etc., and an original / translated sentence storage unit 2 stores an original sentence and a translation as a translation result in association with each other.

Reference numeral 3 denotes a morphological analysis unit that performs morphological analysis on the original sentence. 4 denotes a dictionary search for searching words registered in a word dictionary 12 (described later) for each morpheme analyzed by the morphological analysis unit 3. Department. Reference numeral 5 denotes a parenthesis recognition unit that recognizes, when a character string in parentheses is the first character of some words existing before the parentheses, as a noun phrase. Reference numeral 6 denotes a multi-part-of-speech resolving unit that determines which part of speech is to be used to analyze a word having a plurality of part-of-speech candidates.

8 is based on a morphological analysis unit 3, a dictionary search unit 4, a parenthesis recognition unit 5, a multiple part-of-speech resolving unit 6, and a syntax analysis unit 7.
A source language analysis unit that analyzes the source language.

Reference numeral 9 denotes a conversion unit that converts the syntax of the source language into the syntax of the target language, and reference numeral 10 denotes a target language generation unit that generates a translation in the target language.

11 is an output means for outputting a translation result, 12
Is a word dictionary storing information on words required for translation, and 13 is a rule dictionary storing information on grammar required for translation.

A translation control unit 14 controls the operation of each function of the source language analysis unit 8, the conversion unit 9, and the target language generation unit 10, and the flow of data stored in the source / translation storage unit 2.

FIG. 2 is a circuit block diagram of the machine translator according to the first embodiment of the present invention, and shows the machine translator of this embodiment by hardware.

In FIG. 2, reference numeral 15 denotes an input device for inputting data, such as a keyboard or a pointing device;
16 is a central processing unit (CPU) for controlling the apparatus by executing a program, 17 is a display for displaying data such as a cathode ray tube display or a liquid crystal display, and 18 is a read-only memory (18) for storing data. ROM), 19 is a random access memory (RAM) for temporarily storing data, 20 is a printing device for printing data, such as a printer or a photosetting machine, 21 is a hard disk for storing data, and 22 is data from a recording medium. And a data bus 23.

The ROM 18 has a program storage area for storing programs executed by the CPU 16.

Although the program executed by the CPU 16 is stored in the ROM 18, the program may be recorded on a recording medium such as a portable CD-ROM and read by the reading device 22. good. By doing so, the present invention can be realized by a general-purpose computer.

Here, the correspondence between FIG. 1 and FIG. 2 will be described. 1 and 2, the input means 1 is realized by a keyboard 15. Original / translated sentence storage unit 2 is RAM1
9 is realized. The word dictionary 12 and the rule dictionary 13 are realized by the hard disk 21. The output unit 11 is realized by the display device 17 or the printing device 20.

The morphological analysis unit 3, dictionary search unit 4, parentheses recognition unit 5, multiple part-of-speech resolving unit 6, syntax analysis unit 7, source language analysis unit 8, conversion unit 9, target language generation unit 10, translation control unit 14
Is realized by the CPU 16 executing a control program stored in a program storage area of the ROM 18 while exchanging data with the ROM 18 and the RAM 19.

Each operation flowchart described below is as follows.
FIG. 3 shows how the CPU 16 executes each program stored in the program storage area of the ROM 18.

The operation of the machine translation apparatus configured as described above will be described below. FIG. 3 is an operation flowchart of the machine translation apparatus according to the first embodiment of the present invention, showing an operation of translating one original sentence. In the present embodiment, it is assumed that the original sentence to be translated is already stored in the original / translated sentence storage unit 2.

As shown in FIG. 3, first, when a translation execution instruction is issued from the input unit 1, the translation control unit 14 takes out a sentence to be translated from the original / translated sentence storage unit 2 and sends it to the morphological analysis unit 3. Send (step 1). Here, the “Security Protocol Council (SPC) Inform
ation ”as the translation target sentence.

The morphological analysis unit 3 performs a morphological analysis of the sentence to be translated and divides the sentence to be translated into individual morphemes (step 2). The result is the analysis result shown in FIG.

The dictionary search unit 4 searches the words registered in the word dictionary 12 for each morpheme analyzed in step 2, and obtains data such as the translation of the words and the type of part of speech (step S2). 3). FIG. 28 shows the result.
In FIG. 28, the information for which the translation information does not exist is a word that does not exist in the dictionary, that is, “SPC” or a left parenthesis “(” or a right parenthesis “)” of a symbol.

When the character string in parentheses is the first character of some words existing before the parentheses, the parenthesis recognition unit 5 treats the part as a noun phrase (step 4). The processing of the parenthesis recognition by the parenthesis recognition unit 5 in step 4 will be described in detail with reference to the flowchart shown in FIG.

FIG. 4 is a flowchart showing the operation of the parentheses recognition unit in the first embodiment of the present invention.
This shows the processing of the parenthesis recognition by.

As shown in FIG. 4, first, in step 41, it is determined whether or not a portion enclosed in parentheses exists in the sentence. If there is a portion surrounded by parentheses, the process proceeds to step 42. Here, since “SPC” exists as a portion surrounded by parentheses, the process proceeds to step 42.

In step 42, it is determined whether or not the part existing in parentheses is a word composed of uppercase letters. This is because if the characters in parentheses are composed of uppercase letters, the abbreviations are more likely. Here, since “SPC” is a word composed of capital letters, the process proceeds to step 43.

In step 43, the character string in parentheses is extracted from the rear end one character at a time. Here, first, the last character “C” of “SPC” is extracted.

Proceeding to step 44, it is determined whether or not the character extracted in step 43 is the same as the first character of the word immediately before the parenthesis. Here, since the immediately preceding word is “Council” and the extracted character is “C”, the process proceeds to step 45.

In step 45, it is determined whether the processing has been completed for all the character strings in parentheses. Here, since the characters "S" and "P" have not been processed yet, the process returns to step 43.

Returning to step 43, the character string in parentheses is extracted from the rear end one character at a time, and the next character "
P "is taken out.

Proceeding to step 44, it is determined whether the character extracted in step 43 is the same as the first character of the word immediately before the parenthesis. Here, the immediately preceding word is “Protocol” and the extracted character is “P”.

In step 45, it is determined whether the processing has been completed for all the character strings in parentheses. Here, since the character “S” has not been processed, the process returns to step 43.

Returning to step 43, the character string in parentheses is extracted one character at a time from the rear end, so that the next character "
S "is taken out.

In step 44, it is determined whether the character extracted in step 43 is the same as the first character of the word immediately before the parenthesis. Here, the immediately preceding word is “Security” and the extracted character is “S”, so that the process proceeds to step 45.

In step 45, it is determined whether the processing has been completed for all the character strings in parentheses. Here, since the processing has been completed for all the character strings, the process proceeds to step 46.

In step 46, the character string "SP" in parentheses
Since the part corresponding to "C" was found to be the "Security Protocol Council", "Security Protocol Counci
The l "part is treated as a noun phrase, and the part in parentheses is structured so as to be understood as parenthesis. As a result, the analysis result shown in FIG. 5 is obtained.

When the processing shown in the flowchart of FIG. 4 is completed, the processing returns to the processing of the flowchart shown in FIG.

The multi-speech resolving unit 6 uses the rules of the rule dictionary 13 to determine which interpretation is to be used for analysis if a certain morpheme has a plurality of parts of speech as a result of the parenthesis recognition in step 4. Determine (step 5). Here, “Protocol” has two interpretations, a verb and a noun. However, by utilizing that the “Security Protocol Council” determined in step 4 that the phrase is a noun phrase, “Protocol” is a noun. It is determined. FIG. 6 shows the result.

After performing the multiple part-of-speech elimination in step 5, the syntax analyzer 7 specifies the syntax of the translation target sentence according to the rules of the rule dictionary 13 (step 6). As a result, as shown in FIG. 7, it is determined that the noun has a continuous syntax.

After performing the syntax analysis in step 6, the conversion unit 9 converts the analyzed target sentence in the source language into a target language format (step 7). The result is shown in FIG.

After performing the conversion in step 7, the target language generator 10 generates a translated sentence in the target language based on the output of the converter 9 (step 8). As a result, as shown in FIG. 9, the translated sentence is “Safety Council (SPC) information”.

The translation control unit 14 stores the translation generated in step 8 in the original / translation storage unit 2 and ends the processing (step 9).

As described above, in this embodiment, “Security P
The translation of "rotocol Council (SPC) Information" was changed from "Safety drafts SPC information" to "Safety Council Information (SPC) Information". You will be able to get the correct translation for the abbreviation.

(Embodiment 2) Hereinafter, a second embodiment of the present invention will be described with reference to the drawings. For the configuration of the machine translation apparatus according to the present embodiment using functional units and the configuration using hardware, refer to FIGS. 1 and 2 shown in the first embodiment.

In this embodiment, the invention shown in the first embodiment is applied to the analysis of the whole sentence, and a case where the sentence shown in FIG. 10 is translated will be considered as an example.

When analyzing the first sentence “Security Protocol Council” of FIG. 10 based on the first embodiment of the present invention,
An output of the morphological analysis as shown in FIG. 11 is obtained. On the other hand, if multiple part-of-speech is resolved, a morphological analysis result as shown in FIG. 12 is obtained. When parsing is performed based on this result, FIG.
The analysis result of No. 3 is obtained, and as the translated sentence, "safety prepares council information" is obtained as shown in FIG.

In the sentence of FIG. 10, "Security" is included in the sentence.
The Protocol Council "is a group of words, and its abbreviation is written as" SPC ", but the knowledge is not available.This embodiment makes this knowledge available. Things.

FIG. 15 is a flowchart of the operation of the machine translation apparatus according to the second embodiment of the present invention, showing the operation of registering parentheses in the parenthesis table. In the present embodiment, the original sentence to be translated (see FIG. 10) is
It is assumed that the original / translated sentence storage unit 2 has already stored the data.

As shown in FIG. 15, first, when there is an instruction to execute translation from the input unit 1, the translation control unit 14 extracts one sentence to be translated from the original / translated sentence storage unit 2 (step 11). Here, "Security Protocol Co
uncil Information ”is extracted.

The parenthesis recognition unit 5 registers a sentence that is recognized as a parenthesis from the extracted one sentence in a parenthesis table held therein (step 12). Step 1
The parentheses registration processing by the parentheses recognition unit 5 in 2 will be described in detail with reference to the flowchart shown in FIG.

As shown in FIG. 16, in step 121, it is determined whether or not there is a portion enclosed in parentheses in the original sentence. Here, "Security Protocol Council
Since "Information" does not include a portion surrounded by parentheses, the process is immediately terminated.

Returning to the flowchart shown in FIG. In step 13, it is determined whether all the sentences have been processed. Since all the sentences have not been processed yet, the process returns to step 11.

The translation control unit 14 extracts the next sentence to be translated from the original / translated sentence storage unit 2 (step 1).
1). Here, the “Security Protocol Council (SPC)
is the organization fordiscussing security protoco
ls. "

The parenthesis recognition unit 5 registers, from the extracted one sentence, a sentence recognized as a parenthesis in a parenthesis table held therein (step 12). Step 1
The parentheses registration processing by the parentheses recognition unit 5 in 2 will be described in detail with reference to the flowchart shown in FIG.

As shown in FIG. 16, in step 121, it is determined whether or not there is a part enclosed in parentheses in the sentence. Here, since “SPC” is a portion enclosed in parentheses, the process proceeds to step 122.

In step 122, it is determined whether or not the words in parentheses are words composed of capital letters. If it is a word composed of uppercase letters in parentheses, the process proceeds to step 123. ”SP
Since "C" is composed of capital letters, the process proceeds to step 123.

In step 123, the character string in parentheses is extracted one character at a time from the rear end. First, the last character "C" of "SPC" is extracted.

At step 124, it is determined whether or not the first character of the word immediately before the parentheses is the character extracted at step 123. Here, since the immediately preceding word is “Council” and the extracted character is “C”, the process proceeds to step 125.

In step 125, it is determined whether all the characters in the parentheses have been processed. Since "S" and "P" still remain, the process returns to step 123.

In step 123, the character string in parentheses is extracted one character at a time from the rear end. The next character "P" is retrieved.

At step 124, it is determined whether or not the first character of the word immediately before the parentheses is the character extracted at step 123. Here, since the immediately preceding word is “Protocol” and the extracted character is “P”, the process proceeds to step 125.

In step 125, it is determined whether all the characters in the parentheses have been processed. Since "S" still remains, the process returns to step 123.

In step 123, the character string in parentheses is extracted from the rear end one character at a time. The next letter "S" is retrieved.

At step 124, it is determined whether or not the first character of the word immediately before the parentheses is the character extracted at step 123. Here, the word immediately before is “Security” and the extracted character is “S”.

At step 125, it is determined whether all the characters in the parentheses have been processed. Since all processes have been performed, the process proceeds to step 126.

At step 126, “SP” in parentheses
Since the part corresponding to "C" was found to be the "Security Protocol Council", "Security Protocol Counci
"l" is registered in the column of parenthesis part corresponding words in the parenthesis table, and "SPC" is registered as a parenthesis word, and as a result, the parenthesis table is in the state shown in FIG.

Next, returning to step 121, it is determined whether there is any other part enclosed in parentheses that has not been processed. Here, since there is no data, the flowchart shown in FIG. 16 ends, and step 12 in FIG. 15 ends.

Returning to the flowchart shown in FIG. In step 13, it is determined whether all the sentences have been processed. Since all sentences have not been processed yet, the process returns to step 11, and thereafter, steps 11 to 13 are repeatedly executed for all the sentences in the sentence.

When all the sentences in the sentence have been processed, the process proceeds to step 14 to start the translation process. As shown in FIG.
The translation control unit 14 extracts one sentence to be translated from the original / translated sentence storage unit 2 (step 14). Here, "Se
curity Protocol Council Information ”is retrieved.

Next, the process proceeds to step 15, where one sentence translation is performed. The one-sentence translation process in step 15 is executed according to the flowchart shown in FIG. Steps 1 to 3 are the same processes as those described in the first embodiment, and thus description thereof is omitted.

Regarding the recognition of the parentheses in step 4, the first
Unlike the embodiment, the processing is performed so that the character string registered in the parenthesis table is treated as a noun phrase. This processing will be described in detail with reference to the flowchart shown in FIG.

As shown in FIG. 17, in step 151, one parenthesis corresponding word is extracted from the parenthesis writing table. Here, "Security Protocol Council" is extracted.

Next, the sentence "Security Proto"
It is determined whether or not the extracted character string "Security Protocol Council" exists in the "Col Council Information" (step 152).

In step 153, “Security Protocol
l Council ”is structured as a noun phrase.
An analysis result as shown in FIG. 19 is obtained.

Next, in step 154, it is determined whether the character string immediately after the sentence is the same as the character string in parentheses in the table. Since the character string immediately after is “Information” and the word in parentheses is “SPC”, the process returns to step 152.

Similarly, steps 152 to 155 are repeated. When all the words in the sentence have been processed, the flow proceeds to step 156.

Steps 151 to 156
Is repeated until all the data in the bracketed table is processed. When there is no more data to be processed, the processing according to this flowchart ends, that is, step 4 in FIG. 3 ends, and as a result, FIG. The analysis results shown are obtained.

Returning to step 5 in FIG.
If the result of the parenthesis recognition in step 4 indicates that there is a possibility that a given morpheme has a plurality of parts of speech, the interpretation based on which interpretation is determined using the rules described in the rule dictionary 13 (step 5). ). Here, “Protocol” has two interpretations, verb and noun, but “Security Protocol”
col Council ”is a noun phrase,“ Protoco
l "is determined to be a noun. The result is shown in FIG.

After performing the multiple part-of-speech elimination in step 5, the syntax analyzer 7 specifies the syntax of the translation target sentence using the rules described in the rule dictionary 13 (step 6). As a result, as shown in FIG. 21, it is determined that the noun has a continuous syntax.

After performing the syntax analysis in step 6, the translation unit 9 converts the translated target sentence in the source language into a target language format (step 7). The result is shown in FIG.

After performing the conversion in step 7, the target language generator 10 generates a translation in the target language based on the output of the converter 9 (step 8). As a result, as shown in FIG. 23, the translated sentence is "safety council council information".

The translation control unit 14 stores the translated sentence generated in step 8 in the original sentence / translated sentence storage unit 2 and ends the processing (step 9).

Thus, step 15 in FIG. 15 is completed, and the process proceeds to step 16. Similarly, steps 14 to 16 are repeated until the processing of all the sentences is completed, and the processing ends.

As described above, in this embodiment, a word corresponding to a bracketed portion existing in a document is treated as a noun phrase and stored in a bracketed table, so that even if there is no bracketed portion in one sentence to be translated, A word corresponding to an abbreviation in parentheses existing in a document can be treated as a noun phrase, and the present invention can be applied to a case where a plurality of sentences are translated.

(Embodiment 3) Hereinafter, a third embodiment of the present invention will be described with reference to the drawings. For the configuration of the machine translation apparatus according to the present embodiment using functional units and the configuration using hardware, refer to FIGS. 1 and 2 shown in the first embodiment.

In the present embodiment, when a translated sentence of the target language is generated based on the analysis result of the source language analyzing unit 8 in the first embodiment, a katakana word is selected as the translated word corresponding to the parentheses. The conversion unit 9 is configured to obtain a more readable translated sentence, which will be described below.

In the second embodiment, the conversion unit 9 in the first embodiment is configured to select a katakana word as a translation of a portion corresponding to the parentheses.

The second embodiment is the same as the first embodiment up to step 6 in the flowchart shown in FIG. 3, and a description thereof will be omitted.

In step 7, the conversion unit 9 is configured to select the katakana word as the translation for the part corresponding to the parentheses. Therefore, the conversion unit 9 outputs the output of the syntax analysis unit shown in FIG. 7 to the dictionary search result shown in FIG. Select Katakana from them. As a result, a conversion result shown in FIG. 24 is obtained.

After performing the conversion in step 7, the target language generator 10 generates a translation in the target language based on the output of the converter 9 (step 8). As a result, as shown in FIG. 25, the translated text is “Security Protocol Council (S
PC) information ".

The translation control unit 14 stores the translation generated in step 8 in the original / translation storage unit 2 and ends the processing (step 9).

It is needless to say that this embodiment can be applied to the second embodiment. As described above, in this embodiment, “Security Protocol Council (SPC) Informatio
In the first embodiment, the translation of “n” is “Security Protocol Council (SPC) information” instead of “Safety Protocol Council (SPC) information”, and a more readable translation can be obtained. . This is because proper nouns and technical terms are generally easier to read when expressed in Katakana.

Although the first, second, and third embodiments have been described in detail above, it is needless to say that the present invention is not limited to the algorithm described above. For example, "the Council of Secure Protocol (CS
It is also possible to easily change to correspond to the case where an article is prepended or a preposition is inserted as in "P)".

[0107]

According to the present invention as described above, claim 1
The invention described in (1) describes an original sentence storage unit that stores an input original sentence, a dividing unit that divides the original sentence stored in the original sentence storage unit into morphemes, and a morpheme that is divided by the dividing unit with reference to a word dictionary. A dictionary search means for searching for a part of speech or a translation, and a bracket part for certifying the word part as a noun phrase when the character string in parentheses is the first character of some words existing before the parentheses. Certification means, division means,
Source analysis means for analyzing the syntax of the original text by referring to the rule dictionary based on the results of the dictionary search means and the parenthesis recognition means, and target language generation for generating a translation of the target language based on the analysis results of the original text analysis means Means, and an output means for outputting the translated sentence generated by the target language generating means. When a symbol indicating an abbreviation is given in parentheses in the input original text, the abbreviation corresponds to the symbol. Since a group of words is treated as a noun phrase, a correct translation result can be obtained.

Further, the invention according to claim 2 is characterized in that, in the original text storage means for storing the input original text, and in all the original texts stored in the original text storage means, the character string in parentheses is more than the text in parentheses. A parenthesis registering means for registering the word part as a noun phrase in a parenthesis table when the character is the first character of some preceding words; and a dividing means for dividing the original sentence stored in the original sentence storage means into morphemes A dictionary search unit for searching for a part of speech or a translated word of the morpheme divided by the division unit with reference to the word dictionary, and a bracket unit for certifying a word registered in the parenthesis table as a noun phrase by the bracket unit registration unit. Based on the result of the certification unit, the division unit, the dictionary search unit, and the bracket unit certification unit, refer to the rule dictionary,
Source text analysis means for analyzing the syntax of the original text, target language generation means for generating a translation of the target language based on the analysis result of the source text analysis means, and output means for outputting a translation generated by the target language generation means; By having a configuration that has
Even if there is no part corresponding to the abbreviation in parentheses in one sentence to be translated, if there is a corresponding part in the sentence, the corresponding part is treated as a noun phrase, so that a more correct translation result can be obtained.

The invention described in claim 3 is the first invention.
Alternatively, in the invention according to claim 2, when the target language generating unit generates a translated sentence of the target language based on the result of the analysis by the original sentence analyzing unit, the target language generating unit generates a translated word corresponding to the parentheses as katakana. With this configuration, the translation corresponding to the abbreviation in parentheses is represented in katakana, so that the operator can obtain a translation that is comfortable.

[Brief description of the drawings]

FIG. 1 is a functional block diagram of a machine translation apparatus according to a first embodiment of the present invention.

FIG. 2 is a circuit block diagram of the machine translation apparatus according to the first embodiment of the present invention.

FIG. 3 is an operation flowchart of the machine translation apparatus according to the first embodiment of the present invention.

FIG. 4 is an operation flowchart of a bracket part recognition unit according to the first embodiment of the present invention;

FIG. 5 is a diagram showing an analysis state after parenthesis recognition in the first embodiment of the present invention.

FIG. 6 is a diagram showing an analysis state after eliminating many parts of speech in the first embodiment of the present invention.

FIG. 7 is a diagram showing an analysis state after parsing according to the first embodiment of the present invention;

FIG. 8 is a diagram showing an analysis state after conversion according to the first embodiment of the present invention.

FIG. 9 is a diagram showing a translated sentence in the first embodiment of the present invention.

FIG. 10 is a diagram showing an original in a second embodiment of the present invention.

FIG. 11 is a diagram showing an analysis state after a conventional dictionary search in the second embodiment of the present invention.

FIG. 12 is a diagram showing an analysis state after the conventional multi-speech elimination in the second embodiment of the present invention

FIG. 13 is a diagram showing an analysis state after a conventional syntax analysis in the second embodiment of the present invention.

FIG. 14 is a diagram showing a conventional translated sentence in the second embodiment of the present invention.

FIG. 15 is an operation flowchart of the machine translation apparatus according to the second embodiment of the present invention.

FIG. 16 is an operation flowchart of a parenthesis recognition means in the second embodiment of the present invention.

FIG. 17 is an operation flowchart of a bracket part recognition unit in the second embodiment of the present invention.

FIG. 18 is an explanatory diagram of a parenthesis table according to the second embodiment of this invention;

FIG. 19 is a diagram showing an analysis state after parentheses are recognized in the second embodiment of the present invention.

FIG. 20 is a diagram showing an analysis state after eliminating many parts of speech according to the second embodiment of the present invention.

FIG. 21 is a diagram showing an analysis state after parsing according to the second embodiment of the present invention;

FIG. 22 is a diagram showing an analysis state after conversion according to the second embodiment of the present invention.

FIG. 23 is a view showing a translated sentence in the second embodiment of the present invention.

FIG. 24 is a diagram showing an analysis state after conversion according to the third embodiment of the present invention.

FIG. 25 is a diagram showing a translated sentence in the third embodiment of the present invention.

FIG. 26 is a diagram showing an input sentence in a conventional example.

FIG. 27 is a diagram showing an analysis state after morphological analysis in a conventional example.

FIG. 28 is a diagram showing a dictionary search result in a conventional example.

FIG. 29 is a diagram showing an analysis state after a dictionary search in a conventional example.

FIG. 30 is a diagram showing an analysis state after eliminating many parts of speech in a conventional example.

FIG. 31 is a diagram showing an analysis state after syntax analysis in a conventional example.

FIG. 32 is a diagram showing a translation in a conventional example.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Input means 2 Original / translated sentence storage part 3 Morphological analysis part 4 Dictionary search part 5 Parenthesis part recognition part (parenthesis writing table) 6 Part-of-speech resolving part 7 Syntax analysis part 8 Source language analysis part 9 Conversion part 10 Target language generation part 11 Output means 12 Word dictionary 13 Rule dictionary 14 Translation control unit 15 Input device 16 Central processing unit (CPU) 17 Display device 18 Read only memory (ROM) 19 Random access memory (RAM) 20 Printing device 21 Hard disk 22 Read Device 23 Data bus

Claims (7)

[Claims] 1. An original sentence storage means for storing an input original sentence, a dividing means for dividing an original sentence stored in the original sentence storage means into morphemes, and a morpheme divided by the dividing means with reference to a word dictionary. A dictionary search means for searching the part of speech or translation of the word, and a bracket for identifying the word part as a noun phrase when the character string in the brackets is the first character of some words existing before the brackets Part recognition means, based on the results of the division means, the dictionary search means and the parenthesis part recognition means, referencing a rule dictionary and analyzing the syntax of the original sentence, and analyzing the results of the original sentence analysis means. A machine translation apparatus comprising: a target language generating unit that generates a target language translation based on the target language; and an output unit that outputs a translation generated by the target language generating unit. 2. An original sentence storage means for storing an input original sentence, and in a whole original sentence stored in the original sentence storage means, a character string in parentheses is included in some words existing before parentheses. A parenthesis registering unit that registers the word part as a noun phrase in a parenthesis table, a dividing unit that divides the original sentence stored in the original sentence storage unit into morphemes, and a word dictionary. A dictionary search unit that searches for a part of speech or a translation of a morpheme divided by the division unit; a bracket unit recognition unit that recognizes a word registered in a parenthesis table as a noun phrase by the bracket unit registration unit; Based on the results of the dividing means, the dictionary search means and the parenthesis recognition means, referencing a rule dictionary and analyzing the syntax of the original sentence; Machine translation apparatus comprising: the target language generating means for generating a translation in the target language by Zui, and output means for outputting the translated sentence produced by the target language generating means. 3. The target language generating means, when generating a translated sentence of the target language based on the result of the analysis by the original text analyzing means, generates a translated word corresponding to the parentheses as katakana. Claim 1 or Claim 2
Machine translation device according to the above. 4. A machine translation method for controlling an apparatus having an input device for inputting data, a storage device for storing data, and a display device for displaying data, wherein an original sentence stored in the storage device is retrieved. First step: Divide the extracted original text into morphemes, analyze the part of speech and translation of each morpheme by referring to the word dictionary and rule dictionary, and find the character string in parentheses before the parentheses A second step of treating the word portion as a noun phrase and analyzing the syntax of the original sentence, and translating the target language sentence based on the analysis result in the second step. A third step of generating, and a fourth step of displaying the translation generated in the third step
And a machine translation method. 5. A machine translation method for controlling a device including an input device for inputting data, a storage device for storing data, and a display device for displaying data, wherein an original sentence stored in the storage device is retrieved. The first step is to analyze all the extracted original texts, and if the character string in parentheses is the first character of some words existing before the parentheses, write the word part in parentheses as a noun phrase The second step of registering in the table, dividing the extracted original sentence into morphemes, referring to the word dictionary and the rule dictionary, and analyzing the part of speech and translation of each morpheme,
The third step of treating the character string registered in the parenthesized table in the second step as a noun phrase and analyzing the syntax of the original sentence, and the translation of the target language based on the analysis result in the third step A fourth step of generating; and a fourth step of displaying the translation generated in the fourth step.
And a machine translation method. 6. A recording medium recording a machine translation program for controlling a device including an input device for inputting data, a storage device for storing data, and a display device for displaying data, wherein the storage device stores the machine translation program. A first step of extracting the extracted original sentence, dividing the extracted original sentence into morphemes, referring to a word dictionary and a rule dictionary, and analyzing the part of speech and translation of each morpheme,
A second step of treating the word portion as a noun phrase and analyzing the syntax of the original sentence when the character string in parentheses is the first character of some words existing before the parentheses; A third step of generating a translation of the target language based on the analysis result in the second step, and a fourth step of displaying the translation generated in the third step
And a recording medium storing a machine translation program. 7. A recording medium recording a machine translation program for controlling a device including an input device for inputting data, a storage device for storing data, and a display device for displaying data, wherein the storage device stores the program. A first step of extracting the extracted original text, and analyzing the entire extracted original text, and when the character string in parentheses is the first character of some words existing before the parentheses, the word part is A second step of registering the parenthesis table as a noun phrase, dividing the extracted original sentence into morphemes, referring to a word dictionary and a rule dictionary, and analyzing the part of speech and translation of each morpheme,
The third step of treating the character string registered in the parenthesized table in the second step as a noun phrase and analyzing the syntax of the original sentence, and the translation of the target language based on the analysis result in the third step A fourth step of generating; and a fourth step of displaying the translation generated in the fourth step.
And a recording medium storing a machine translation program.