JP5039114B2 - Machine translation apparatus and program - Google Patents

Description

  The present invention relates to a machine translation apparatus and a program for translating an original sentence of a first language to be translated into a translated sentence of a second language using an example of translation consisting of a parallel translation of a first language sentence and a second language sentence.

  With the progress of internationalization, machine translation devices are highly expected as tools for realizing rapid information exchange using foreign languages. At present, the results of translation by the current machine translation technology cannot be said to be at a level that does not require any manual retouching. Therefore, there is a translation support tool that can be used to create a database of translation results that have been translated manually. Has been developed. In these tools, a plurality of translation examples composed of a translation of the first language source text and the second language translation text are stored in advance in the translation example database, and the source text is similar to the input source text to be translated. Search and display translation examples (hereinafter referred to as similar examples). If the original text of the retrieved similar example does not completely match the original text to be translated, a part of the translated text of the retrieved translation example is edited to obtain a translation result of the original text to be translated. .

  In the machine translation apparatus using such a translation example, there is also proposed a machine translation apparatus that automatically performs the editing in order to reduce the user's trouble of determining an edit location and selecting a translation (for example, Patent Document 1). reference). In this case, a difference (hereinafter, difference) between the original text to be translated and the retrieved original text of the similar example is determined in terms of words, and in the similar example, the phrase in the translation corresponding to the difference in the original text is determined, and the phrase A synthesized sentence in which is translated into a translation corresponding to the difference in the translation target original is created and used as a translation of the translation target original.

JP 2006-11842 A

However, the difference in the similar example assumed in Patent Document 1 is assumed to be a single word (glasses, suitcase, etc.), but does not include the case where the synthesized word corresponds to the difference. Is the current situation. In practice, a compound word composed of a plurality of words may correspond to a difference, and in such a case, determination of a phrase in the translation of the similar example corresponding to the difference in the original sentence to be translated may not be successful. . For example, an example shown in Table 1 will be described. The example of Table 1 is a case where the original text to be translated is translated using the translation example (similar example original sentence, similar example translation).

  In this example, it is assumed that the difference between the original text to be translated and the original text of the similar example searched is “remote”, and even if an attempt is made to find a word in the translation corresponding to “remote” in the original text in the similar example, There was a problem that a word directly corresponding to "remote" could not be found. Since “remote treatment” in this example is a compound word composed of “remote” and “treatment”, there is a one-to-one correspondence relationship between “remote”, “treatment”, and “teletherapy”. Not being judged is the cause of the problem. As a result, the phrase in the translation corresponding to the difference in the original sentence cannot be found, and the replacement of the translation has failed.

  As described above, in the translation using the similar example, in the conventional technique, the difference between the original text to be translated and the similar example original is determined in units of words, and the phrase of the similar example translation corresponding to the difference in the original of the similar example is determined. In this case, there is a case where the determination of the word or phrase in the translation of the similar example corresponding one-to-one with the difference in the original sentence fails.

  An object of the present invention is to provide a machine translation apparatus and a program capable of determining the range of corresponding translation words in a translation even when the difference between the translation target text and the translation example text is a compound word not included in the translation example translation The purpose is to do.

A machine translation apparatus according to the present invention includes a machine translation program, a machine translation dictionary for translating a first language to be translated into a second language for translation, a sentence in the first language to be translated and a second language for translation. A translation example consisting of parallel translations with a similar example original sentence and a translation example database stored as a pair of similar example translations, an input device for inputting sentence data of a first language to be translated, and a translation A machine translation apparatus comprising: an output device that outputs a translated sentence in a later second language; and an arithmetic control device that executes the machine translation program. The sentence data of the first language to be translated input from the input device An input processing unit that accepts and stores it in the storage device, a sentence division unit that divides sentence data of the first language received by the input processing unit into one sentence unit, and obtains a translation target original sentence, and the sentence division A source sentence analysis / translation unit that analyzes the translation target text obtained in step 1 with reference to the machine translation dictionary and translates it into a second language, a translation example similar to the translation target text, and a similar example translation a translation example search unit searches from the translation example database as pairs of the similar example together corresponding to the difference between the determined difference between the translation example search section the similar example textual retrieved in said translated original text A difference associating unit for associating a phrase in a translated sentence with a phrase in the original text to be translated, a phrase determined as a difference in the difference associating unit, and a phrase that failed to be matched one-to-one in similar examples A re-association determination unit that expands the range of each word and re-associates, and a difference word in the original text to be translated that is associated by the difference association unit A differential word / phrase translation acquisition unit that acquires a translation in a corresponding second language with reference to the machine translation dictionary, and acquires a difference phrase / phrase translation of a phrase in a similar example translation translated by the difference correlation unit A translation replacement unit that generates a translation by replacing the translation acquired by the translation unit.

  According to the present invention, it is possible to determine the range of corresponding translation words in a translation even if the difference between the translation target sentence and the translation example original is a compound word that is not included in the translation example translation.

The functional block diagram of the machine translation apparatus concerning embodiment of this invention. The block block diagram which shows the hardware constitutions of the machine translation apparatus concerning embodiment of this invention. The flowchart which shows the processing content of the machine translation apparatus concerning embodiment of this invention. FIG. 4 is a tree structure diagram illustrating an example of an analysis result obtained by performing language analysis processing in step S305 of FIG. 3 on the translation target original text illustrated in Table 1. The flowchart which shows the content of the difference matching process including the matching retry process of step S308 of FIG. FIG. 6 is a tree structure diagram illustrating an example of an analysis result obtained by performing language analysis processing in step S401 of FIG. 5 on the similar example original document shown in Table 1. FIG. 6 is a tree structure diagram showing an example of an analysis result obtained by performing language analysis processing in step S402 of FIG. 5 on the similar example translations shown in Table 1.

  Embodiments of the present invention will be described below. FIG. 1 is a functional block diagram of a machine translation apparatus 11 according to the embodiment of the present invention, and FIG. 2 is a block configuration diagram showing a hardware configuration of the machine translation apparatus according to the embodiment of the present invention.

  First, the hardware configuration of the machine translation apparatus according to the embodiment of the present invention will be described.

  In FIG. 2, the machine translation device 11 is realized 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.

  Next, a functional configuration of the machine translation apparatus 11 according to the embodiment of the present invention will be described. In FIG. 1, each functional block in the arithmetic control device 12 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 input processing unit 26 receives text data in the first language input from the input device 20 and stores it in the storage device 25. For example, when text data is directly input by the user from the keyboard 22 or the like, or is input from the disk drive 23 via a storage medium, the text data is input into the memory 14 or the storage device 25 in the arithmetic control device 12. To remember.

  The sentence division unit 27 divides the sentence data of the first language received by the input processing unit 26 and stored in the storage device 25 into sentence units, and stores them in the work area 16 of the memory 14 as original sentences to be translated. The original text divided in units of one sentence is transferred to the control unit 28.

  The control unit 28 includes an original text analysis / translation unit 29, a translation example search unit 30, a difference association unit 31, a reassociation determination unit 32, a difference word / phrase translation acquisition unit 33, a translation replacement unit 34, a machine translation dictionary 35, a translation The example database 36 and the output processing unit 37 are controlled.

  The source text analysis / translation unit 29 analyzes the text with reference to the machine translation dictionary 35 with respect to the source text to be translated, which is input from the input processing unit 26 and divided into single sentences by the text division unit 27, Translates.

  The translation example search unit 30 uses the translation target original text divided and input by the sentence splitting unit 27 as a search key as a search key from the translation example database 36 and uses the similar example original sentence and the example translation sentence. Search as a pair.

  The difference associating unit 31 associates the difference between the searched similar example original and the translation target original based on the analysis result by the original sentence analyzing / translating unit 29, and translates the words and phrases in the similar example translated sentence corresponding to the difference. This is associated with a word or phrase in the target original text.

  The reassociation determination unit 32 determines the difference association when the difference association unit 31 fails to determine a phrase in the similar example translation that corresponds one-to-one to the difference between the similar example original and the translation target text. For the words in the similar example source text that the part has determined to be the difference, the words in the source text to be translated, and the words that failed to be matched one-to-one in the similar example translation, expand the range of each word and respond again It is a thing to do.

  The difference word / phrase translation acquisition unit 33 acquires a translation in the second language corresponding to the difference in the original text specified by the difference association unit 31 or the reassociation determination unit 32 with reference to the machine translation dictionary 35. is there.

  The translation replacement unit 34 generates a translation by replacing the phrase in the similar example translation specified by the difference association unit 31 or the reassociation determination unit 32 with the translation acquired by the difference phrase translation acquisition unit 33. is there.

  The machine translation dictionary 35 stores vocabulary and rules necessary for analyzing the input original text in the first language and translating it into the second language, and is formed in the storage device 25. Vocabulary / rules necessary for translation are stored in the vocabulary part 35a, morpheme analysis rule part 35b, syntax analysis rule part 35c, conversion rule part 35d, and generation rule part 35e of the machine translation dictionary 35. Similar examples are stored in the translation example database 36.

  The output processing unit 37 outputs the result of translation generated by the original text analysis / translation unit 29 to the output device 17. For example, display is output to the display device 13 as the output device 17. The output device 17 is not limited to the output from the display device 13 but may be output to the audio output device by voice in addition to the output from the display device 13 or in addition to the output from the display device 13 or may be output to the printing device. Alternatively, it may be output to a recording medium via the disk drive 23.

  Next, the contents of the translation processing in the machine translation apparatus 11 according to the embodiment of the present invention will be described using Japanese-English translation as an example. FIG. 3 is a flowchart showing the processing contents of the machine translation apparatus according to the embodiment of the present invention.

  When the user inputs sentence data of the first language to be translated from the input device 20, first, the input processing unit 26 receives the sentence data to be translated (S301). The input sentence data to be translated is stored in the memory 14 or the storage device 25.

  Next, the sentence division unit 27 reads the sentence data to be translated from the memory 14 or the storage device 25 and divides it into one sentence unit (S302). For the translation target text that is one sentence of the divided text data to be translated, the control unit 28 sets “1” to the variable I (S303), and extracts the I th translation target text (S304). The I-th original text to be translated is taken out to the work area 16 of the memory 14.

  Next, the original sentence analysis / translation unit 29 performs a language analysis process with reference to the machine translation dictionary 35 with respect to the taken-out original sentence to be translated (S305). That is, the source sentence analysis / translation unit 29 performs morpheme analysis processing of the translation target original sentence with reference to the vocabulary part 35a and the morpheme analysis rule part 35b of the machine translation dictionary 35, and performs syntax analysis processing with reference to the syntax analysis rule part 35c. To obtain information on the part of speech and syntax of each word constituting the original text to be translated. The parsing process and other techniques used here are not specific to the present invention, but are general techniques related to language processing.

  FIG. 4 is a tree structure diagram illustrating an example of an analysis result obtained by performing the language analysis processing in step S305 of FIG. 3 on the translation target original text shown in Table 1. Words and phrases surrounded by solid lines are each element constituting a sentence obtained by morphological analysis, and a syntactic connection between the elements is expressed by a line connecting these elements. A diagram showing the syntactical connection between the words constituting the sentence is called a tree structure.

  Subsequently, the translation example search unit 30 searches the translation example database 36 for a translation example that is the same as or similar to the translation target original text using the translation target original text as a search key (S306). The search for similar examples is performed in consideration of the degree of matching of sentences calculated from the ratio of words that match each word constituting the original text to be translated, the matching of other attributes, and the like.

  Note that in determining whether words match, it is possible to perform processing in which the fluctuation of the notation is not regarded as a difference, using the result of the language processing acquired in step S305. The notation of katakana refers to katakana notation such as the presence or absence of a long clef symbol, the shaking of a kana notation, the shaking of kanji and hiragana notation, the shaking of the presence or absence of a kana. Of course, it is also possible to employ a configuration in which only completely matching translation examples that do not absorb the fluctuation of the notation are used as search results. In addition, even if there is a difference between characters and words, if the difference ratio is less than a specific value, the difference is ignored. The structure to control may be sufficient. This makes it possible to extract translation examples that are similar in terms and phrases. In this case, the timing of the example search (step S306) need not be after the source language analysis (step S305). This is because the result of dictionary lookup / morpheme analysis is not required.

  In the example search in the translation example search unit 30, a plurality of similar examples may be searched, but the example having the highest similarity is finally selected as a target for narrowing down processing. In order to determine an example having the highest priority from a plurality of similar examples, for example, when the similarity and the similarity are the same, it is generally performed using criteria such as the registrant of the example and the registration date and time. is there.

  The control unit 28 determines whether or not a similar example has been searched (S307). When the similar example is searched by the translation example searching unit 30, the control unit 28 stores the similar example in the similar example storage area of the memory 14. Remember. Subsequently, the difference association unit 31 and the reassociation determination unit 32 perform difference association processing including association retry processing on the similar example to be processed (S308).

  That is, in step S308, first, the difference associating unit 31 performs difference associating processing on the similar example (similar example original sentence and similar example translated sentence) and the translation target original sentence. When the determination of the words in the similar example translation that corresponds one-to-one with the difference between the words fails, the re-association determination unit 32 performs the one-to-one association between the words determined to be the difference and the similar examples. The range of failed words is expanded and the association retry process is performed. Details of the processing in step S308 will be described later.

  For the difference word / phrase in the original text to be translated determined by the difference association unit 31 or the re-association determination unit 32, the difference word / phrase translation acquisition unit 33 refers to the vocabulary part 35a of the machine translation dictionary 35 and translates the translated word. A difference word / phrase translation acquisition process is performed (S309). Thereafter, the translation replacement unit 34 uses the translation acquired by the differential phrase translation acquisition unit 33 and uses the translation in the similar example translation corresponding to the differential phrase determined by the differential association unit 31 or the reassociation determination unit 32. A translated word replacement process is performed (S310). Then, the translation obtained as a result of the translation replacement process is taken as the translation result of the original translation target sentence.

  On the other hand, if it is determined in step S307 that there is no similar example, the source text analysis / translation unit 29 performs translation processing of the source text to be translated (S311). That is, the process related to the difference is not performed (S308 to S310), and the process proceeds to step S311. The original sentence analysis / translation unit 29 performs the translation process of the I-th original sentence (S311).

  Then, the output processing unit 37 displays the translation result obtained by the processing of step S310 or step S311 (S312). That is, the output processing unit 37 displays the translation result on the display device 17. Regarding the content displayed on the display device 17, if no similar example is detected in step S307, the translation result obtained in step S311 is displayed. In step S307, a similar example is found, and in the difference association process including the association retry process in step S308, the phrase in the translation of the similar example corresponding one-to-one to the difference in the original text of the similar example is successfully determined. If it is, the translated text obtained by replacing the translation corresponding to the difference of the translation target original text obtained in step S309 is displayed for the similar example translated text.

  Although a similar example was detected in step S307, in the difference association process including the association retry process in step S308, the phrase in the similar example translation that corresponds one-to-one with the difference in the similar example original sentence is determined. If it has failed, the similar example translation detected in step S307 is displayed as a reference translation.

  Then, it is determined whether or not processing has been completed for all text data to be translated, that is, whether or not all text processing has been completed (S313). If processing has been completed for all text data to be translated, processing is performed. When the processing is finished and the processing for all the original sentences is not finished, “1” is added to the variable I (I = I + 1) (S314), and the process returns to step S304.

  FIG. 5 is a flowchart showing the contents of the difference association process including the association retry process in step S308 of FIG. The original sentence analysis / translation unit 29 takes out the similar example original sentence and the translation target original sentence into the work area 16 of the memory 14, and performs a language analysis process on the similar example original sentence acquired by the translation example search part 30 (S401). .

  FIG. 6 is a tree structure diagram showing an example of an analysis result obtained by performing the language analysis processing in step S401 of FIG. 5 on the similar example original text shown in Table 1. Words and phrases surrounded by solid lines are each element constituting a sentence obtained by morphological analysis, and a syntactic connection between the elements is expressed by a line connecting these elements. The words surrounded by a broken line are a list of candidate words registered in the vocabulary part 35a as second word candidate words for each element constituting the similar example original.

  Similarly, the source text analysis / translation unit 29 performs language analysis processing on the similar example translation (S402). FIG. 7 is a tree structure diagram showing an example of an analysis result obtained by performing the language analysis processing in step S402 of FIG. 5 on the similar example translations shown in Table 1. Words and phrases surrounded by solid lines are each element constituting a sentence obtained by morphological analysis, and a syntactic connection between the elements is expressed by a line connecting these elements. The words surrounded by a broken line are a list of candidate words registered in the vocabulary part 35a as candidate words of the first language for each element constituting the similar example translation. Items with multiple translations are listed separated by commas, and those with no translation are represented by “Φ”.

  The difference association unit 31 uses the language analysis result (FIG. 4) obtained in step S305 of FIG. 3 and the language analysis result (FIG. 6) obtained in step S401 of FIG. And the difference from the similar example original is determined (S403). This determines the difference between the translation target original and the similar example original by matching the tree structure between FIG. 4 and FIG. Thereby, in the case of the example of Table 1, a difference is determined like the following [1]. This is a difference determined from character string information.

[1] Original text to be translated: (N / A)-Similar example Original text: Adjective noun “remote”
The difference associating unit 31 associates the phrase in the similar example original text corresponding to the difference with the phrase in the similar example translation (S404). This is done with reference to the tree structure of FIGS. That is, the words of the similar example original sentence and the similar example translated sentence are associated by the following [A]. The difference associating unit 31 performs the difference associating process for each constituent element of the similar example original (S404).

[A-1] For a phrase (first language) constituting a similar example original sentence, a search is made for a match from the translated words (first language) of each word constituting the similar example translation (second language), and a match is found. The constituent word of the similar example translation sentence having the translation word to be extracted is extracted as a corresponding word candidate (second language) of the similar example translation sentence corresponding to the constituent word of the similar example original sentence. If there are a plurality of candidates, a plurality of candidates are extracted.

[A-2] For each word (second language) constituting the similar example translation, search for a match from the translation (second language) of each word constituting the similar example original (first language), The constituent words of the similar example original sentence having the matching translation are extracted as corresponding word candidates (first language) of the similar example original sentence corresponding to the constituent words of the similar example translated sentence. If there are a plurality of candidates, a plurality of candidates are extracted.

  Then, the difference association unit 31 determines whether or not the words in the similar example original text and the words in the similar example translation sentence have a one-to-one correspondence (S405). The phrase is determined as the corresponding phrase, and thereby the correspondence between the difference and the phrase of the similar example translation sentence is determined (S406).

  In the case of the example in Table 1, “Use” in the similar example source text is “use” in the similar example source text, “Medical device” in the similar example source text is “medical device” in the similar example source text, “This” in the similar example source text. ”Corresponds to the similar example translation“ this ”, the similar example original“ cerebral hemorrhage ”corresponds to the similar example translation“ cerebral hemorrhage ”, and the similar example original“ teletherapy ”corresponds to the similar example translation“ teletherapy ”. Accordingly, since the difference “remote” in the similar example original text cannot be correlated one-to-one with the words in the similar example translation, the one-to-one correlation has failed in the determination in step S405.

  If the difference association unit 31 fails to associate one-to-one with the words in the similar example translation of the difference, can the reassociation determination unit 32 expand the range of the words to be associated? It is determined whether or not the corresponding range can be expanded (S407).

  The re-association determining unit 32 tries to expand the range of the words to be associated with each other by the following method when the one-to-one correspondence is not determined. Whether or not the range of correspondence can be expanded is determined by whether or not the following range expansion enabling condition [B] is satisfied.

[B] Range expandable condition: a phrase that is continuous with the detected phrase on the structure tree, and has the same part of speech attribute as the detected phrase, or a part of speech that can modify the detected phrase Shall.

  In the example of Table 1, “remote” as the search target is continuous with “treatment” and “intracerebral hemorrhage” on the structure tree, and has the same part-of-speech attribute of “noun”. It is determined that 32 satisfies the range expansion enabling condition.

  When the re-association determination unit 32 determines that the range expansion enable condition is satisfied, the re-association determination unit 32 expands the corresponding range (S408). The range expansion is performed under the following conditions.

[B-1] Search for partially matching words (first language) constituting the similar example original sentence from the translated words (first language) of each word constituting the similar example translated sentence (second language) Then, an unmatched portion of the found translated word is cut out, and the constituent elements of the similar example original sentence having the portion as the translated word are searched.

[B-2] With respect to a phrase constituting the similar example translation (second language), a search is performed for a partial match from the translation (second language) of each phrase constituting the similar example original (first language). Then, an unmatched portion of the found translated word is cut out, and the constituent elements of the similar example original sentence having the portion as the translated word are searched.

  In the example shown in Table 1, the component “treatment” of the similar example original text and the “teletherapy” which is the translated word of the component “teletherapy” of the similar example translation match the “treatment” portion. In this case, “remote”, which does not match “remote treatment”, is cut out, and a search is performed as to whether or not a similar example original sentence component having “remote” as a translated word exists.

  Alternatively, in the tree structure of the original text of the similar example, the part of speech or the word that modifies “treatment” that is connected to “treatment” and has the same attribute as “treatment” is connected, and the resultant composite word is “ It is determined whether or not it matches “teletherapy”.

  As a result, “distant”, which is a word (adjective noun) that can modify “treatment” having “tele-” as a translation, is detected. Finally, the corresponding range is expanded from “treatment” in the original text of similar examples to “remote” + “treatment”.

  Then, the process returns to step S408, and the difference associating unit 31 sets the difference as [2] below based on the similar example original “remote” + “treatment” expanded by the re-association determining unit 32. judge.

[2] Original text to be translated: noun "treatment" --- similar example
In the example of Table 1, when the word determined to be a difference is a word that modifies a noun, another candidate for difference [2] whose range has been expanded to a modified noun is also determined as a difference. Accordingly, the “remote treatment” of the similar example original text and the “teletherapy” of the similar example translation text are associated with each other in the determination in step S404 of the difference associating unit 31. In other words, the following correspondence is obtained between the words constituting the similar example original sentence and the similar example translated sentence by the above association processing.

Use --use
Medical device--medical device
This --this
Cerebral hemorrhage
Teletherapy--teletherapy
In step S405, it is determined that the phrase of the similar example original sentence and the phrase of the similar example translation have a one-to-one correspondence, and the associated phrase is determined as the corresponding phrase. As a result, the correspondence between the difference “treatment” and the phrase “teletherapy” of the similar example translation is determined (S406). In addition, since the conditions as described above are set, for example, even if the modifier “therapeutic” is added to the “medical device” in the similar example original text, the search is performed because it exists in a position far away on the tree structure. Not subject.

  Using the above results, the difference word / phrase translation acquisition unit 33 determines that the following one-to-one correspondence is obtained among the translation target original sentence, the similar example original sentence, and the similar example translated sentence. In addition, since the one-to-one correspondence between the three parties cannot be obtained in the difference [1], the difference [2] is adopted in this case.

Original text "Treatment"-Similar example Original text "Teletherapy"-Similar example translation "teletherapy"
Using the above determination results, the difference word / phrase translation acquisition unit 33 acquires the translation of “treatment”, which is the difference of the translation target original text, with reference to the vocabulary unit 35a (S309). As described in FIG. 6, candidate words for “treatment” registered in the vocabulary part 35 a include “treatment” and “therapy”. Of these candidate translations, the word “teletherapy” in the similar example translation corresponding to the difference phrase of the similar example translation is replaced by the translation replacement unit 34 using “treatment” registered first (S310). Thereby, in the example of Table 1, similar example translations as shown in Table 2 are obtained.

  Next, as a condition for expanding the range of a word / phrase, in addition to the above-described range expansion enabling condition [B], a condition “limit character type” may be set. That is, it is possible to set a condition that the character type is the same as the word from which the range is expanded, or is composed of a character type that does not include hiragana (consisting only of katakana, alphabet, and kanji). Under this condition, an unknown word that is not registered in the vocabulary part 35a can be targeted for range expansion. For example, the phrase “BB Agency” exists in the original text of the similar example, and “BB” is not registered in the Japanese-English dictionary of the vocabulary part 35a but is registered in the English-Japanese dictionary (“BB device” "BB agency" is registered as a translation of ")", the word type does not contain hiragana even if the part of speech of the original "BB" is unknown. Can do.

  As described above, in the embodiment of the present invention, the difference between the translation target text and the similar example original text is determined, and the one-to-one correspondence between the phrase in the similar example translation and the determined difference fails. In this case, the association process can be retried. By performing the retry, the probability of association failure can be reduced.

  When retrying, the range of differences is expanded based on specific conditions related to parts of speech, etc., for the words before and after the word determined to be a difference, so the correspondence between the original sentence of the similar example and each word constituting the translated sentence Among the candidate candidates, there is a syntactic connection, and the candidates that can be regarded as one compound word can be accurately narrowed down. Also, an unknown word that is not registered in the machine translation dictionary of machine translation, and even if an attribute such as a part of speech is unknown, the word can be expanded.

  In the embodiment of the present invention, Japanese-English translation has been described. However, it is applicable not only to Japanese-English translation but also to English-Japanese translation and translation between other languages, such as Chinese-Japanese translation and Japanese-Chinese translation. Needless to say.

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 ... Input processing unit, 27 ... Sentence division unit, 28 ... Control unit, 29 ... Original text analysis / translation unit, DESCRIPTION OF SYMBOLS 30 ... Translation example search part, 31 ... Difference matching part, 32 ... Re-association possible judgment part, 33 ... Difference word phrase translation acquisition part, 34 ... Translation word replacement part, 35 ... Machine translation dictionary, 35a ... Vocabulary part, 35b ... Morphological analysis rule part, 35c ... Syntax analysis rule part,
35d ... Conversion rule part, 35e ... Generation rule part, 36 ... Example database for translation, 37 ... Output processing part

Claims (4)

Machine translation program, machine translation dictionary for translating a first language to be translated into a second language for translation, translation example comprising a translation of a sentence in the first language to be translated and a sentence in the second language to be translated A storage device storing a translation example database that stores a similar example original and a similar example translation as a pair, an input device that inputs a translation target original, an output device that outputs a translated second language translation, and the machine In a machine translation device comprising a calculation control device for calculating and executing a translation program, a source text analysis / analysis for analyzing a source text input from the input device with reference to the machine translation dictionary and translating into a second language A translation example, a translation example search unit that searches the translation example database as a pair of a translation example of a similar example and a translation example of a similar example, and the source solution - based on the analysis result obtained by the translation unit, the phrase of the similar example in translation together corresponding to the difference between the determined difference between the Translation example search section the similar example textual retrieved in said translated original text A one-to-one correspondence between a difference association unit that associates words in a translation target original, a word in a similar example original determined as a difference in the difference association unit, a phrase in a translation target original, and a similar example translation The re-association determination unit that expands the range of each word and re-associates with the words that failed, and the translation associated with the difference association unit and the re-association determination unit A difference word / phrase translation acquisition unit that acquires a translation in a second language corresponding to a difference word / phrase in the target original text by referring to the machine translation dictionary, the difference association unit, and the reassociation Possible determination unit machine translation apparatus is characterized in that a translation replacement unit for generating a translated sentence by replacing the word similar example in translation associated with the translation obtained by the difference word translation acquisition unit by. The re-association determination unit expands the range of the word as a condition for re-association, and is continuous with the word on the structure tree and has the same part-of-speech attribute as the word or the word 2. The machine translation apparatus according to claim 1, wherein: The re-association determination unit, as a condition for re-associating by expanding the range of the word, the word is continuous on the structure tree and the character type is the same as the word or includes hiragana 2. The machine translation apparatus according to claim 1, wherein the machine translation apparatus is provided on the condition that it is composed of no character type. Machine translation program, machine translation dictionary for translating a first language to be translated into a second language for translation, translation example comprising a translation of a sentence in the first language to be translated and a sentence in the second language to be translated A storage device storing a translation example database that stores a similar example original and a similar example translation as a pair, an input device that inputs a translation target original, an output device that outputs a translated second language translation, and the machine In a computer for causing a computer to function as a machine translation device having a computation control device for computing and executing a translation program, the computer analyzes the original text to be translated input from the input device with reference to the machine translation dictionary and The translation example database as a pair of a translation example similar to the translation target sentence and a similar example original sentence and a similar example translation sentence. A function to search, on the basis of the translated textual analysis results, the phrase of the similar example in translation together corresponding to the difference between determining the difference between the the retrieved the similar example original and the translated original text A function for associating a phrase in the original text to be translated with a phrase in the similar example original sentence determined to be the difference, a phrase in the original text to be translated, and a phrase that failed to correspond one-to-one in the similar example translation sentence A function for expanding the range of each word and re-associating, and referring to the machine translation dictionary for the corresponding difference word in the original text to be translated, and acquiring a translation in the corresponding second language And a function for generating a translation by replacing a word / phrase in a corresponding similar example translation with the acquired translation.

Images