JP2010287149A - Bilingual document proofreading apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a bilingual document proofreading apparatus that determines whether proofreading is suitable even when both first language document and second language document have been proofread. <P>SOLUTION: A document analysis means 28 analyzes an unproofread first language document, an unproofread second language document, a proofread first language document and a proofread second language document. A proofreading classification means 30 compares the proofread first language document with the unproofread first language document and the proofread second language document with the unproofread second language document according to the analysis information to classify proofreadings into proofreading classes. A proofreading suitability determination means 32 determines whether the proofreadings are suitable according to the analysis information and proofreading classes. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a bilingual document proofreading apparatus that proofreads a bilingual document of a first language document before proofreading and a second language document before proofreading.

  With the progress of globalization, there is a growing need for translating various documents into English and other foreign languages, and vice versa. With the development of natural language processing technology, machine translation technology has also improved. Documents that need to be translated are superior to manual translation in terms of cost and speed, and use of machine translation has increased. On the other hand, manual intervention is still necessary to make the translation result of machine translation an acceptable document. This is especially true for documents where translation quality is particularly important.

  Documents that need to be translated are often requested to be translated at the initial draft stage of the manuscript to be distributed to readers whose native language is a foreign language almost simultaneously with the distribution of the original document. While the translator is translating the first draft, the original author often reviews the manuscript and increases the difference from the first draft. When the final draft was completed, the translator compared the first draft with the final draft to determine what the difference between them was, and detected the corresponding portions of the first draft and made necessary corrections. This work put a great burden on the translator.

  In the case of manual translation, it is indispensable to have knowledge of both the language of the original document and the language for translation as a minimum requirement, but the amount of knowledge is rarely the same, and in most cases , Either is dominant. In other words, when a Japanese document is translated into English, it is often done by a person whose native language is Japanese and English is a foreign language. For this reason, there is often a native check.

  However, this native check is not without problems. First, the worker who performs the native check may have knowledge of the source language or may not have knowledge of the source language. If the operator does not have knowledge of the original language, the operator does not refer to the original document and proofreads only by looking at the translated document. Therefore, there is no guarantee that the proofreading result is in accordance with the purpose of the original document. In addition, there are places where native checks are defined as such.

  For example, in http://www.eigo-honyaku.jp/term/process/nativecheck.html, “The person who performs the native check is called the native checker. "There is no question about the expertise of the translation."

  Secondly, even if the worker has some knowledge of the source language, the reality is as mentioned in http://777ch.info/. The translated text is scanned without going through the reconciliation check, and the excessively terrible part is reworked on the surface, and even if it is repaired well on the surface, there is a serious defect inside. If you write well on the surface, it becomes even more difficult to notice problems such as mistranslations. "

  For this reason, in order to make the best use of the effects of native checking, it is necessary for a professional translator who is familiar with both languages to examine the proofreading results. Although this case is also of a different type, it is required to refer to three types of documents as in the first case. In other words, the translation document before the native check, the translation document reflecting the native check, and the original document. Therefore, this confirmation work places a great burden on the operator.

  When such a translation check operation is performed, when the original text and the translated text are displayed on the display means, when the corrected original text corrected by the input means is input, the corrected original text is compared with the original text. Then, it is determined whether or not the corrected original corresponds to the reworking of the original, and when it is determined that the corrected original corresponds to the reworking of the original, the corrected translated text in which the corrected original is translated and the translated text in which the original is translated are determined. In comparison, the difference between the two is identified and displayed so that the difference between the translated text and the corrected translated text can be easily recognized, thereby improving the efficiency of the translation work (for example, see Patent Document 1).

  In other words, when there is an undesired translation, the corresponding original sentence is handled instead of the translation, and the desired translation is adopted by machine translation again. If the changes made to the original text correspond to the “rework” criteria that the inventor says, the difference between the machine translation results before and after the change is identified and displayed, and the operator adopts either . When it is determined that the result is not similar to “rework” and is not similar, the machine translation result of the original after the change is displayed as it is. The judgment criteria for reworking are as follows, for example.

(1) The number of characters that differ between the original text and the corrected original text or between the translated text and the corrected translated text is equal to or less than a predetermined number.

(2) Characters that differ between the original text and the corrected original text or between the translated text and the corrected translated text are only differences in notation.

(3) When the original sentence and the corrected original sentence, or the translated sentence and the corrected translated sentence are divided by words, only the order is different.

(4) When the original text and the corrected original text, or the translated text and the corrected translated text are divided by words, different words are stored in the dictionary as synonyms.

(5) When the original sentence and the corrected original sentence, or the translated sentence and the corrected translated sentence are divided into words, the number of different words is equal to or less than a predetermined number.

  In the case of Patent Document 1, when a change is made to a translation, a change is made to the translation of the machine translation, and it is determined that the translation after the change corresponds to “rework” The machine translation program having a reverse translation function is used to translate the translation before the change and the translation after the change into the original language, and the difference between the two is identified and displayed. The operator selects either one.

  In addition, the second natural language machine translation sentence machine-translated from the first natural language input sentence is automatically proofread, and the translated sentence after the operator proofreads the result of machine translation of the spoken language. Is compared with the translation of machine translation, the proofreading rule is automatically derived from it, and the proofreading rule is automatically applied by applying the generated proofreading rule to the proofreading sentence (for example, patent Reference 2).

  In addition, the translation is obtained from the parallel translation data that has the original similar to the untranslated original, and the translation is compared with the original to be translated and the similar original and only the difference is translated. There are some which have been designed to be realized (see, for example, Patent Document 3).

JP 2006-178682 A Japanese Patent No. 3387437 JP 11-272671 A

  However, the conventional version deals with the case of creating a higher-quality translation by proofreading the result of machine translation of the original text to be translated. We did not judge the validity of the proofreading performed.

  Therefore, the operator compares the original text or translation before and after the proofreading with the original text or translation after the proofreading, detects the difference, and the corrected text or translation is valid for all the differences. It was necessary to determine whether it was present, and to make necessary changes to the original or translated text as necessary, which entailed significant work costs.

  An object of the present invention is to provide a bilingual document proofreading apparatus that can determine the suitability of proofreading contents even when both a first language document and a second language document are proofread.

  The bilingual document proofreading apparatus according to the present invention relates to a storage device storing a bilingual document proofreading program and a translation dictionary unit, a first language document and a second language document as a result of the translation. Input device for inputting previous second language document, first language document after proofreading and second language document after proofreading, as well as information necessary for proofreading of bilingual documents A bilingual document proofreading apparatus comprising a display device for performing the bilingual document proofreading program, and a first language document before proofreading and a second language document before proofreading input from the input device, And a bilingual document storage unit that stores a proofreading first language document and a proofreading second language document as a pair and stores them as a bilingual document, and is stored in the bilingual document storage unit using the translation dictionary unit. Before calibration Document analysis means for analyzing one language document, second language document before proofreading, first language document after proofreading and second language document after proofreading, and storing the analysis information in an analysis information storage unit, and the analysis information storage Compare the first language document before proofreading with the first language document after proofreading, and the second language document before proofreading with the second language document after proofreading based on the analysis information stored in the The calibration content classification means for storing the calibration classification in the calibration classification storage unit, the calibration information based on the analysis information stored in the analysis information storage unit and the calibration classification stored in the calibration classification storage unit Calibration content suitability judging means for judging the suitability of the content and displaying the calibration content suitability result on a display device, and prompting the operator to confirm the calibration content suitability result displayed on the display device and Through the input device Characterized by comprising a calibration contents confirmation operation means for performing a confirmation operation of the calibration details of the force has been operated first language document and after calibration of the second language document after calibration based on the command.

  ADVANTAGE OF THE INVENTION According to this invention, even if it is a case where both the 1st language document and the 2nd language document are proofread, the bilingual document proofreading apparatus which can judge the suitability of the proofreading content can be provided.

1 is a configuration diagram of a bilingual document proofing apparatus according to an embodiment of the present invention. The block block diagram which shows the hardware constitutions of the bilingual document proofreading apparatus which concerns on embodiment of this invention. The flowchart which shows an example of the processing content of the bilingual document proofreading apparatus concerning embodiment of this invention. Explanatory drawing of an example of the bilingual document before proofreading in embodiment of this invention. Explanatory drawing of an example of the bilingual document after proofreading in embodiment of this invention. Explanatory drawing which shows an example of the morphological analysis information of the first sentence of the 1st language document before proofreading among the bilingual documents before proofreading in embodiment of this invention. Explanatory drawing which shows an example of the morphological analysis information of the first sentence of the 2nd language document before proofreading among the bilingual documents before proofreading in embodiment of this invention. Explanatory drawing of an example of the calibration classification memorize | stored in the calibration classification memory | storage part in embodiment of this invention. The flowchart which shows an example when the proofreading content suitability determination means in embodiment of this invention performs the judgment content suitability determination process in the same language. FIG. 10 is an explanatory diagram of an example of a bilingual document in the case of FIG. 9. Explanatory drawing which shows the example of analysis of the sentence contained in the bilingual document of case 4 of FIG. The flowchart which shows another example in the case of performing the judgment process of the proof content suitability in the same language by the proof content suitability judging means in the embodiment of the present invention. The flowchart which shows an example in the case of performing the judgment process of the proofreading content suitability in the different language of the bilingual document before and behind proofreading by the proofreading content suitability judging means in embodiment of this invention. FIG. 14 is an explanatory diagram of an example of a bilingual document in the case of FIG. 13. Explanatory drawing which shows an example of the morphological analysis information of the 1st language document among the bilingual documents before proofreading of FIG. Explanatory drawing which shows an example of the morphological analysis information of the 2nd language document among the bilingual documents before proofreading of FIG. Explanatory drawing which shows an example of the morphological analysis information of the 2nd language document among the bilingual documents after proofreading of FIG. The flowchart which shows another example at the time of performing the judgment process of the proofreading content suitability in the different language of the bilingual document before and behind proofreading by the proofreading content suitability judging means in embodiment of this invention. The flowchart which shows an example in the case of performing the judgment process of the proof content suitability in case the calibration classification | category is a combination or addition of deletion and addition by the proofreading content suitability judging means in the embodiment of the present invention. Explanatory drawing of an example of the bilingual document in the case of applying FIG. The flowchart which shows the processing content of the calibration content confirmation operation means at the time of a user determining calibration content suitability. FIG. 22 is an explanatory diagram of an example of a bilingual document when FIG. 21 is applied. Explanatory drawing of an example of the matrix of proofreading content approval.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a configuration diagram of a bilingual document proofreading apparatus according to an embodiment of the present invention, and FIG. 2 is a block configuration diagram illustrating a hardware configuration of the bilingual document proofreading apparatus according to an embodiment of the present invention. The bilingual document proofreading apparatus 11 is realized by installing a software program such as a bilingual document proofreading program in a general computer and executing the software program in the processor 13 of the arithmetic and control unit 12.

  The arithmetic and control unit 12 performs various calculations related to bilingual document proofreading. The arithmetic and control unit 12 includes a processor 13 and a memory 14. The memory 14 stores a bilingual document proofreading program 15 related to translation. The work area 16 is used when processing is executed. The calculation results of the calculation control device 12 are displayed and output on the display device 17 which is an output device.

  The input device 18 is used to input information to the arithmetic control device 12, and includes, for example, a mouse 19, a keyboard 20, and a disk drive 21. For example, the mouse 19 and the keyboard 20 are connected to the arithmetic control device 12 via the display device 17. Various commands are input to the keyboard 20 and the disk drive 21 to input a document to be proofread for bilingual documents. In other words, the disk drive 21 inputs / outputs a document file to be proofreaded into a bilingual document to / from a storage medium. Furthermore, a hard disk drive (HDD) 22 is provided for storing a translation dictionary and the like in which calculation results of the calculation control device 12 and knowledge / rules necessary for bilingual document proofreading are accumulated.

  FIG. 1 is a functional block diagram of a bilingual document proofreading apparatus 11 according to an embodiment of the present invention. Each functional block in the arithmetic and control unit 12 shown in FIG. 1 corresponds to each program constituting the bilingual document proofreading program 15 described above. That is, when the processor 13 executes each program constituting the parallel translation document proofing 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 22 in the arithmetic control device 12.

  In FIG. 1, an input device 18 inputs electronic data of a document to be proofread. The input device 18 includes, for example, a keyboard 20, a mouse 19 and the like, and can input a bilingual document based on a user input operation. is there. A document input from the input device 18 includes a first language document before proofreading and a bilingual document before proofreading in which the first language document before proofreading is translated into a second language. This is a bilingual document after manual proofreading for this pre-proofread bilingual document. The proofread bilingual document is a bilingual document in which the proofread first language document and the proofread second language document are paired. As the input device 18, a reading device from a computer-readable medium such as an OCR (optical character reading device), a magnetic tape, a magnetic disk, an optical disk or the like can be adopted. The pre-proofread bilingual document and the post-proofread bilingual document are input and captured by the input processing unit 23 of the arithmetic processing unit 12 and stored in the bilingual document storage unit 26 of the storage device 25 via the control unit 24. The input device 18 gives various commands to the control unit 24 via the input processing unit 23.

  The control unit 24 controls the input processing unit 23, the output processing unit 33, the document analysis unit 28, the proofreading content classification unit 30, the proofreading content suitability determination unit 32, and the proofreading content check operation unit 34, and data with the storage device 25. Also controls the transfer of.

  The document analysis unit 28 uses the translation dictionary unit 27 of the storage device 25 to be described later according to an instruction from the control unit 24 and is input by the input device 18 and stored in the parallel translation document storage unit 26 before the first language. The document, the second language document before proofreading, the first language document after proofreading, and the second language document after proofreading are analyzed, and the analysis information is stored in the analysis information storage unit 29. The analysis result is displayed and output on the display device 17 through the control unit 24 and the output processing unit 33 as necessary.

  Based on the analysis information stored in the analysis information storage unit 29, the proofreading content classification means 30 is based on the first language document before proofreading and the first language document after proofreading, and the second language document before proofreading and after proofreading. The proofreading contents are compared with the second language document, the proofreading contents are classified, and the proofreading classification is stored in the proofreading classification storage unit 31.

  The calibration content suitability determination means 32 judges the suitability of the calibration content based on the analysis information stored in the analysis information storage unit 29 and the calibration classification stored in the calibration classification storage unit 31, and controls the calibration content suitability result. The data is output to the display device 17 via the unit 24 and the output processing unit 33.

  The calibration content confirmation operating means 34 prompts the user to confirm the calibration content suitability result displayed on the display device 17, and based on an operation command input by the user via the input device 18, An operation for confirming the proofreading contents of the one language document or the second language document after proofreading is performed.

  The output processing unit 33 outputs the bilingual proofreading information such as the bilingual document, the analysis result, the proofreading classification, the proofreading content suitability result supplied through the control unit 24 to the display device 17. The bilingual proofreading information is displayed on the 17 display screen. The output processing unit 33 displays responses from the control unit 24 to various commands to the control unit 24.

  Although the display device 17 is shown as an output device, the output device is not limited to the display device 17, but may be a printing device such as a printing machine, an output device to a computer-readable medium such as a magnetic tape, a magnetic disk, or an optical disk. It is also possible to employ a transmission device that transmits a document to other media.

  The translation dictionary unit 27 stores various types of dictionary data used when the proofreading content aptitude determination means 32 performs the proofreading content aptitude determination process. The translation dictionary unit 27 stores a dictionary for performing translation from the first language to the second language and a dictionary for performing translation from the second language to the first language.

  The dictionary for translating from the first language to the second language is a first language utilization change dictionary 27a for converting words / idioms whose endings are changed into their original forms, and for analyzing the first language. A first language analysis grammar dictionary 27b in which grammar is stored, a first language word / idiom dictionary 27c in which translations in a second language corresponding to words / idioms in the first language are stored together with the part of speech information, A first language conversion grammar dictionary 27d that stores conversion information into two languages, a second language generation grammar dictionary 27e that determines the structure of a sentence in the second language, and further changes the word form such as endings to complete a translated sentence. Consists of a second language morpheme generation grammar dictionary 27f.

  A dictionary for translating from the second language to the first language is a second language utilization change dictionary 27g, a second language analysis grammar dictionary 27h in which a grammar for analyzing the second language is stored, and a second Second language word / idiom dictionary 27i in which the translation of the first language corresponding to the word / idiom of the language is stored together with the part of speech information, and the second language conversion in which the conversion information from the second language to the first language is stored It comprises a grammar dictionary 27j, a first language generation grammar dictionary 27k that determines the structure of a sentence in the first language, and a first language morpheme generation grammar dictionary 27l that completes a translated sentence by changing word forms such as endings. Here are some useful dictionaries, but not all need to be used.

  In the following, description will be made taking as an example the case where English is the first language and Japanese is the second language. FIG. 3 is a flowchart showing an example of processing contents of the parallel document proofreading apparatus according to the embodiment of the present invention. First, the control unit 24 stores the bilingual document before proofreading and the bilingual document after proofreading input from the input device 18 via the input processing unit 23 (S1). That is, the first language document before proofreading and the second language document before proofreading are used as bilingual documents before proofreading, and the first language document after proofreading and second language documents before proofreading are used as bilingual documents after proofreading. And stored in the bilingual document storage unit 26.

  FIG. 4 is an explanatory diagram of an example of a bilingual document before proofreading, and FIG. 5 is an explanatory diagram of an example of a bilingual document after proofreading. The bilingual document in FIGS. 4 and 5 shows a case where the second language document is a verbatim translation in which the words of the first language document are translated without excess or deficiency, and the first language document of the bilingual document is two sentences. Yes. Each sentence is given a sentence number.

  In the bilingual document after proofreading in FIG. 5, the sentence number may not necessarily be one sentence. For example, sentence number 2 of the second language document after proofreading in FIG. 5 is not one sentence but two sentences. This is because the sentence number here indicates a sentence corresponding to one sentence in the first language before proofreading. In other words, in the bilingual document after proofreading, there is no sentence when sentences are deleted or combined by proofreading, and there may be two or more sentences when sentences are added by proofreading. Moreover, even in the same sentence before proofreading, the same thing can be said because one sentence in the first language is not necessarily one sentence in the second language. Actually, the second language sentence corresponding to sentence number 2 in FIG. 4 is two sentences. In addition, different portions are underlined for easy understanding in the description of the comparison processing described later.

  Here, there is already software for determining what is to be inserted / deleted / replaced from the surface character string, as described in, for example, WER (Word Error Rate). This reflects the edit distance (insertion / deletion / replacement) compared to the reference exemplary translation.

  Next, the control unit 24 activates the document analysis means 28. The document analysis means 28 reads the parallel translation document before proofreading and the parallel translation document after proofreading from the parallel translation document storage unit 26, divides each sentence of the parallel translation document before proofreading and the parallel translation document after proofreading into morphemes. Attribute information is obtained (S2). This is to obtain syntactic features of each sentence of the bilingual document before proofreading and the bilingual document after proofreading.

  FIG. 6 is an explanatory diagram showing an example of morphological analysis information of the first sentence of the first language document before proofreading among the bilingual documents before proofreading. The first sentence of the first language document before proofreading is a dictionary for translation from the first language to the second language in the translation dictionary unit 27, specifically, the first language utilization change dictionary 27a and the first language. By matching with the analysis grammar dictionary 27b, as shown in FIG. 6, a part of speech, a prototype, and an attribute are given to each word.

  FIG. 7 is an explanatory diagram showing an example of morphological analysis information of the first sentence of the second language document before proofreading among the bilingual documents before proofreading. Japanese, which is the second language, is an agglutinative word, so unlike the refraction word, English, the boundaries of the words are not visually understood, and the translation dictionary unit 27 translates from the first language to the second language. The morpheme analysis is performed using the dictionary for use, specifically the second language utilization change dictionary 27g and the first language analysis grammar dictionary 27h. Thereby, in the case of the first sentence of the second language document of FIG. 4, for example, it is divided as shown in FIG. 7, and a part of speech, a prototype, and an attribute are given to each word.

  Next, using the first language word / idiom dictionary 27c and the second language word / idiom dictionary 27i, translation information defined in the translation dictionary unit 27 is obtained for each morpheme (S3). In the case of the first sentence on the first language side, compound words such as “scientific community,” and “human genome” are also registered as headings in the first language word / idiom dictionary 27c. Words are also subject to translation information. The same processing is performed for the second sentence of the bilingual document before proofreading. The same processing is also performed on the proofread bilingual document. The morpheme analysis information and the translated word information obtained here are stored in the analysis information storage unit 29 by the control unit 24.

  Next, the control unit 24 activates the calibration content classification means 30. The proofreading content classification means 30 reads the bilingual document before correction and the bilingual document after proofreading from the bilingual document storage unit 26, and determines whether or not the sentences with the same sentence number in the bilingual document before proofreading and the bilingual document after proofreading are the same. Judgment is made (S4). That is, the sentence with the same sentence number in the first language document before proofreading and the first language document after proofreading, and the sentence with the same sentence number in the second language document before proofreading and the second language document after proofreading are the same. Judge whether or not. For the same sentences, the proofreading content classification means 30 assigns the same classification as the proofreading classification (S5). Then, the calibration content classification unit 30 stores the calibration classification in the calibration classification storage unit 31 via the control unit 24.

  FIG. 8 is an explanatory diagram of an example of calibration classification stored in the calibration classification storage unit 31. In the bilingual document before proofreading in FIG. 4 and the bilingual document after proofreading in FIG. 5, the first sentence (sentence number 1) and the second sentence (sentence number 2) of the first language document before proofreading are after proofreading, respectively. Since the first sentence (sentence number 1) and the second sentence of the first language document are the same, they are classified as the same as shown in FIG.

  On the other hand, for the sentences that are not identical, the proofreading content classification means 30 determines whether or not there is a partially overlapping part in the sentence having the same sentence number (S6). If there is no overlapping part, it is determined whether or not the same sentence exists in a sentence with a different sentence number (S7). If there is the same sentence, sentence movement is given as a proofreading classification (S8). Whether the same sentence exists in a sentence with a different sentence number is determined by checking some neighboring sentences in the document in the same language after proofreading.

  If it is determined in step S6 that the same sentence does not appear in another sentence number, it is determined whether or not the sentence that is not identical is a sentence before proofreading (S9). If it is a sentence before proofreading, a sentence deletion classification is assigned as a proofreading classification (S10). On the other hand, if a sentence that is not identical is a proofreading sentence, a sentence addition classification is assigned as a proofreading classification (S11). ).

  On the other hand, if it is determined in step S6 that there is a partial overlap, the proofreading content classification unit 30 first determines whether a phenomenon of sentence division or sentence combination is observed (S12). As a criterion for this determination, the content words are the same except for the overlapping function words, and it is determined whether or not the number of sentences is different before and after the proofreading.

  Whether it is a content word or a function word can be obtained by referring to the part-of-speech information of the morphological analysis information shown in FIGS. As parts of speech, nouns, adjectives, verbs, and adverbs belong to content words, pronouns, prepositions (in English), conjunctions, and interjections belong to function words. That is, the proofreading content classification unit 30 reads out the morphological analysis information of each sentence of the parallel translation document before the proofreading and the parallel translation document after the proofreading from the analysis information storage unit 29 and refers to the part of speech information of the morphological analysis information, thereby It is determined whether or not the content words match except for the function words. If the content words are the same except for the function words in the overlapping portion and the number of sentences is different before and after proofreading, it is determined that a phenomenon of sentence division or sentence combination is observed.

  Next, it is determined whether or not the number of sentences is reduced after proofreading (S13). When the number of sentences is reduced after proofreading, a sentence combination classification is assigned as a proofreading classification (S14). On the other hand, if the number of sentences has increased after proofreading, sentence division classification is assigned as proofreading classification (S15).

  For example, in the second sentence of the second language in FIG. 5 “The National Institutes of Health has officially started a preliminary study. This is to create a comprehensive inventory of cancer-related genomic changes. “The National Institutes of Health officially launched a preliminary study to create a comprehensive inventory of cancer-related genomic alterations, in the second sentence of the second language of FIG. Is divided into two sentences, a sentence division classification is assigned.

  On the other hand, if it is determined in step S12 that the phenomenon of sentence division or sentence combination is not observed in the partially overlapping portion, the proofreading content classification unit 30 determines whether the document is a parallel translation document before proofreading. If it is a bilingual document before proofreading (S16), a deletion classification is given as a proofreading classification (S17). On the other hand, if it is not a bilingual document before proofreading, it is a bilingual document after proofreading, so an additional classification is given as a proofreading classification (S18).

  For example, the first sentence “Smith of Smith” in the second language document in FIG. 4 is classified as deleted, and the first sentence “Scientist like Smith” in the second language document in FIG. 5 is added. being classified. In addition, “biology” of “cancer ecology” that existed before proofreading is classified as deleted because it does not exist in the sentence after proofreading. The second sentence of the second language document in FIG. 5 is “that is,“ cancer genome atlas ”next to the sentence division described above. This exists before and after calibration. However, after proofreading, “abbreviated TCGA is named after four types of bases” follows, which is different. Therefore, this sentence is classified as additional.

  Then, as shown in FIG. 8, the proofreading content classifying unit 30 collects the proofreading classifications assigned for each sentence number of the parallel translation document (S19).

  Next, based on this proofreading classification, the proofreading content aptitude determination means 34 determines whether the same sentence classification is used on the first language side and the second language side for the same sentence number (S20). If they are the same, it is determined that there is a high possibility of having proofreading content suitability (S21). On the other hand, if it is different, it is determined whether it is any of sentence movement, sentence combination, and sentence division (S22). If it is any of sentence movement, sentence combination, and sentence division, the proofreading content is suitable. It is determined that the possibility is high (S21). If it is not any of sentence movement, sentence combination, and sentence division, it is determined that the possibility of having proofreading content suitability is low (S23). This is because sentence movement, sentence combination, and sentence division are only paraphrasing, and there is no content addition or deletion, so it is considered that there is a possibility of having proofreading content suitability.

  Here, in the case of the bilingual documents of FIGS. 4 and 5, the first sentence is the same before and after proofreading on the first language side, whereas the part before proofreading is deleted on the second language side. There is an addition in the same place after calibration. In other words, there is a replacement. Furthermore, in the subsequent part, the part that existed before the calibration is deleted, and there is no addition to the deleted part. Therefore, the proofreading contents suitability determination means 34 determines that the proofreading contents suitability may be low.

  The second sentence is the same before and after proofreading on the first language side, while there are sentence division and addition on the second language side. Since the sentence combination has proofreading content aptitude, it becomes a comparison between “same” and “addition”. Accordingly, since the second sentence has a different proofreading classification, the proofreading content suitability determining unit 34 determines that the proofreading content suitability may be low.

  On the other hand, there are cases where there is a high possibility that the proofreading content is likely to be suitable for the first language side, when proofreading and additional editing are performed after proofreading and after proofreading, as well as sentence movement, sentence combination, sentence division Is when only can be seen.

  Thus, when all the sentence numbers are determined, the proofreading content confirmation operation means 34 displays the proofreading content suitability result on the display device (S24). Then, it prompts the user for confirmation operation and determines whether or not there is an operation command from the user (S25). When there is an operation command from the user, the calibration content confirmation operation means 34 follows the operation command. The process is completed (S26). In this case, the proofreading content confirmation operation means 34 can also exclude the proofreading contents determined by the proofreading content aptitude determination means 32 from being subjected to the check correction work. Thereby, the confirmation correction work by a user can be performed rapidly.

  As described above, the locations where the calibration has been performed are classified into “things that are likely to have low suitability for calibration contents” and “things that are likely to have suitability for calibration contents”. Since “things that may have low suitability for calibration content” require visual confirmation, the location corresponding to this classification is highlighted, for example, to make it easier for the operator. You can also.

  In the above description, the suitability of the proofreading contents is judged using both the first language and the second language of the bilingual document before and after the proofreading. However, the bilingual document before and after the proofreading is judged within the same language. May be. Specifically, the case where the character string that existed before the proofreading is deleted and a new character string is added at the position of the character string that was deleted after the proofreading, that is, the determination for replacement is performed.

  FIG. 9 is a flowchart showing an example of the case where the proofreading content suitability determination means 32 performs the proofreading content suitability determination process in the same language before and after the proofreading, and FIG. 10 is an explanatory diagram of an example of the bilingual document in that case. . In the following description, the case where the same language of the bilingual document before and after proofreading is English of the first language will be described.

  The proofreading content aptitude determination means 32 reads the proofreading classification of the target bilingual document from the proofreading classification storage unit 31 (S1). In this case, the target bilingual document is each sentence of the first language document of the bilingual document shown in FIG. Then, it is determined whether or not the calibration classification is deleted or added (S2).

  If the proofreading category is not deleted or added, the process is terminated. If the proofreading category is deleted or added, the proofreading content aptitude determination unit 32 reads the analysis information of the target bilingual document from the analysis document storage unit 26 and reads the part of speech. And so on (S3).

  Here, in case 1 (sentence number 1) in FIG. 10, it can be seen that the before-correction grasp is a transitive verb of the original grasp past tense, and the suggested after the correction is the original suggest past tense. In case 2 (sentence number 2), both the first before proofreading and the initial after proofreading are adjectives.

  On the other hand, Case 3 (sentence number 3) covers a plurality of words, and is 3 words before and after proofreading. “Patience with customers” before proofreading consists of a noun plural of patience: noun, with: preposition, customers: original customer. “In sales networking” after proofreading consists of in: preposition, sales: adjective, networking: noun. The former can be expressed as [patience [with customers] PP] NP and the latter can be expressed as [in [sales networking] NP] PP, using the generation grammar notation method. The latter is a prepositional phrase, and the part of speech composition is different.

  In case 4 [statement number 4], “there was” before calibration and “he held” after calibration are different. From the part of speech, there are: adverb, was: be verb past tense, he: pronoun, held: transitive hold past tense.

  FIG. 11 is an explanatory view showing an analysis example of a sentence included in a bilingual document of case 4 (sentence number 4), FIG. 11 (a) is an analysis example of a sentence after proofreading, and FIG. This is an example of sentence analysis. As shown in FIG. 11, none of them constitutes a constituent as a two-word block. In FIG. 11, an element at a node can be a constituent element. In other words, “held a breakfast meeting with potential contributors” is a component, but “held a” and “he held” are not components. Therefore, looking at the position of the sentence as a whole, it can be seen that although it is the same two words, the part of speech is different, and therefore the role that other invariant parts occupy in the sentence is significantly different. That is, the sentence before proofreading is the there syntax, and there is a fixed part. Subsequent “a breakfast meeting with potential contributors” refers to the existence itself. The role of the sentence is the subject. On the other hand, in the sentence after proofreading, he held is the head of the verb phrase that is the subject and predicate of the main sentence, respectively. In this sentence, the same “a breakfast meeting with potential contributors” is the object of this other verb held.

  Next, it is determined whether or not the different parts of speech are the same (S4). If they are different from each other, it is determined that the possibility of having the calibration content suitability is low (S5), and if they are the same, it is determined that the possibility of having the calibration content suitability is high (S6).

  Since case 1 (sentence number 1) and case 2 (sentence number 2) in FIG. 10 have the same part of speech, it is determined that there is a high possibility of having proofreading content suitability, and case 3 (sentence number 3) in FIG. Since the part of speech is different from Case 4 (Sentence No. 4), it is determined that the possibility of having proofreading content suitability is low. Thus, when the proofreading classification is deletion or addition, if the part of speech of the deleted word matches the part of speech of the word added by replacement of the deleted word, it is determined that the proofreading content is suitable. .

  In FIG. 9, when the part of speech of the deleted word matches the part of speech of the word added by replacement of the deleted word, it is determined that the proofreading content is suitable, but as shown in FIG. Alternatively, it may be determined that each word has suitability for proofreading contents when the translation dictionary portion of the corresponding language is dictionary-didged and they are synonyms. In this case, there is a higher possibility that the proofreading content is suitable.

  In FIG. 12, steps S4 to S5 are the same as those in FIG. If the parts of speech of the different parts are the same, it is determined whether or not the words of the different parts are synonyms (S6). This determination is performed by lexicographically searching each word for the corresponding language translation dictionary and determining whether or not they are synonyms. If it is a synonym, it is determined that there is a high possibility of having proofreading content suitability.

  That is, when the parts of speech are the same, focus on the meanings of the different words. The first language word / idiom dictionary 27c and the first language word / idiom dictionary 27i of the translation dictionary unit 27 also have synonym information. Therefore, the proofreading content aptitude determination means 32 uses the first language word / idiom dictionary 27c as a synonym of first, “archetypal, archetypical, prototypal, prototypic, prototypical, basic, introductory, initial, firstborn, eldest, freshman , first-year, original, premier, premiere, prime, prime ”, etc., it is determined that first and initial are synonymous because there is an initial after proofreading.

  Since initial is included as a synonym for first, the investigation may be terminated for the synonym of initial, or bidirectional equivalence may be confirmed. Examining initial as well, the synonym for the adjective initial is "first". Thus, strong support is obtained in which the adjectives first and initial are synonymous with each other.

  On the other hand, let's look at grasp and suggest in case 2 (sentence number 2). Similarly, using the first language word / idiom dictionary 27c, the synonyms of grasp are examined, and the words “hold on, hold, take hold, get the picture, comprehend, savvy, dig, compass, apprehend, understand” are obtained. It is done. It can be seen that this does not include suggestions. Then, when the synonym of suggest is checked next, for example, the following words are obtained. "Propose, advise, declare, hint, intimate, imply, indicate, inform, evoke, paint a picture, express, show, evince" is obtained, and this does not include the grasp. If there is no synonym as in case 2 (sentence number 2), it is determined that the possibility of proofreading content suitability is low (S5). If there is a grasp, it is determined that there is a high possibility of having the proofreading content suitability.

  Here, the case of replacement with different notation was not considered, but if the first language word / idiom dictionary 27c and the second language word / idiom dictionary 27i contain information related to different notation, the above description will be given. If “synonym” is replaced with “different notation”, the suitability of the proofreading content can be similarly determined. Further, although it is determined in step S4 whether or not the parts of speech are the same, even if they are not exactly the same, it is possible to determine an allowable range in advance and make a flexible determination.

  Furthermore, according to the above, the locations where calibration was performed were classified as “low possibility of having calibration content suitability” and “high possibility of having calibration content suitability”. Since “a low possibility of having” needs to be visually confirmed, it is possible to make the operator's convenience by making a display device such as highlighting a portion corresponding to this classification.

  Next, in FIG. 9 and FIG. 12, in determining the suitability of the proofreading content, the bilingual document before and after the proofreading is judged in the same language. You may make it perform.

  Specifically, the case where the character string that was before proofreading was deleted on both the first language side and the second language side, and a new character string was added to the part of the character string deleted after proofreading, that is, Make a replacement decision.

  FIG. 13 is a flowchart showing an example of a case where the proofreading content suitability determination means 32 performs a proofreading content suitability determination process in a different language before and after the proofreading, and FIG. 14 is an explanatory diagram of an example of the bilingual document in that case. .

  The proofreading content aptitude determination means 32 reads the proofreading classification of the target bilingual document from the proofreading classification storage unit 31 (S1). In this case, the target bilingual document is each sentence of the first language document of the bilingual document shown in FIG. Then, it is determined whether or not the calibration classification is deleted or added (S2).

  If the proofreading category is not deleted or added, the process is terminated. If the proofreading category is deleted or added, the proofreading content aptitude determination unit 32 reads the analysis information of the target bilingual document from the analysis document storage unit 26 and reads the part of speech. And so on (S3).

  FIG. 15 is an explanatory diagram showing an example of morphological analysis information of a first language document in a bilingual document before proofreading, and FIG. 16 is an explanatory diagram showing an example of morphological analysis information of a second language document in a bilingual document before proofreading. FIG. 17 is an explanatory diagram showing an example of morpheme analysis information of the second language document in the proofread bilingual document.

  The sentence in the first language of the bilingual document before proofreading shown in FIG. 14 is composed of words as shown in FIG. 15, and parts of speech, prototypes, and attributes are assigned to the words. The sentence structure of the first language sentence of the bilingual document is the SVO sentence pattern of the subject, the transitive verb, and the object.

  On the other hand, the sentence in the second language of the bilingual document before proofreading is composed of words as shown in FIG. 16, and each part has a part of speech, a prototype, and an attribute. The sentence structure of the second language sentence before proofreading is a passive voice composed of a subject and a verb. The deep structure is an SOV sentence pattern, and it is thought that it has been transformed into passive, but at that time S is lost.

  The sentence in the second language of the bilingual document after proofreading is composed of words as shown in FIG. 17, and each part has a part of speech, a prototype, and an attribute. The part-of-speech structure is as shown in FIG. 17, and the sentence structure of the sentence is the subject, object, and transitive SOV sentence pattern.

  Next, it is determined whether or not the sentence structure of the first language document before proofreading and the second language document after proofreading have a correspondence in terms of sentence pattern (S4).

  In other words, in step S4, the different part of the second language after proofreading with respect to the second language before correction “scholars have presented many explanations so far”, which corresponds to the original text “Scholars have proposed many explanations” To compare. As you can see from the above, “scholars have presented many explanations so far” consists of the main part “scholars” + the adverb “previously” that modifies the predicate + the object and the transitive verb It has an SOV structure of predicate “I have presented many explanations”, and “Scholars have proposed many explanations” is an SVO called “have proposed many explanations” consisting of the main part “scholars” + transitive verb and object. Has a structure.

  Therefore, the sentence structure of the first language document before proofreading is SVO, and the sentence structure of the second language document after proofreading is SOV. Considering whether or not there is a correspondence in sentence pattern, it is determined as follows.

  For example, regarding the sentence pattern correspondence between English and Japanese, English is the right-branching “right branch” language, and Japanese is left-branching “left branch”. The word precedes the verb, but in Japanese the object follows the verb. Considering this linguistic property, “SVO” in English and “SOV” in Japanese are treated as having a sentence pattern correspondence.

  In general, the languages of the world are roughly SOV type (subject + object + verb), VSO type (verb + subject + object), OSV type (object + subject + verb), VOS type (verb + object + Subject) and OVS type (object + verb + subject). The classification of each language is stored in advance as information, and the difference in word order is absorbed based on this information.

  Now, if the sentence before and after proofreading in the second language is reversed in FIG. 14, the different part of the noticed “Many arguments have been made about the evolution of primate intelligence”. "There has been much discussion so far." This is passive, and it can be said that the first language “Scholars have proposed many explanations” and “scholars have presented many explanations so far” are different in that they are active. Accordingly, in this case, it is determined that the possibility of having the proofreading content suitability is low, and the process is terminated.

  Next, it is determined whether or not the different parts of speech are the same (S5). If they are different from each other, it is determined that the possibility of having the calibration content suitability is low (S6), and if they are the same, it is determined that the possibility of having the calibration content suitability is high (S7).

  FIG. 13 illustrates the case where the sentence structure of the first language document before proofreading and the sentence structure of the second language document after proofreading are determined and processed in a sentence pattern. The same applies to the case where the sentence structure of the second language document and the sentence structure of the first language document after proofreading are processed by determining whether or not the sentence structure has a correspondence relationship.

  Further, in FIG. 13, when the part of speech of the deleted word matches the part of speech of the word added by replacement of the deleted word, it is determined that the proofreading content is suitable, but as shown in FIG. Further, it is determined whether or not there is a word in a different language among the translation candidates of the word added by the replacement, and when there is a word in a different language, it is determined that there is a high possibility that the proofreading content is suitable. It is a thing. The same steps as those in FIG. 13 are denoted by the same reference numerals, and redundant description is omitted.

  In FIG. 18, step S8 is added to FIG. In step S8, according to the determination in step S4, "the scholars have presented many explanations so far about the evolution of primate intelligence" in the sentence of the second language after the proofreading. Now that we know that it is at least structurally compatible with the language document, we next try to make a judgment based on meaning.

  Of the different parts, "scholars have presented many explanations so far." In this sentence, since “explanation” is the head of the object, it can be predicted from a structural viewpoint that the word corresponding to the head of the object corresponds to “explanation” even in the corresponding sentence in the first language before proofreading.

  Therefore, “explanation” is compared with “explanation”. The first language word / idiom dictionary 27c of the translation dictionary unit 27 has Japanese translation information with English as an entry word. From this, Japanese translation candidates for explanation are listed as follows. "Explanation, interpretation, clarification, explanation, explanation, excuse, meaning, interpretation, discussion, reconciliation, collusion, mutual concession." Since there is an “explanation” in this list, it is determined that correspondence has been taken.

  Here, because there is “explanation” among the Japanese translation candidates for explanation, you may discontinue the survey on the degree of coincidence between “explanation” and “explanation”, but here you should also investigate from the opposite direction To. The second language word / idiom dictionary 27i has English translation information with Japanese as a headword. Therefore, enumeration of English explanation candidates for “explanation” includes “account, explanation, exposition, interpretation, elucidation, illustration, presentation, representation, portrayal, description, caption, legend”, etc. Also includes “explanation”.

  Therefore, it is understood that “explanation” is a valid translation of “explanation”. If this semantic correspondence is different and at least the main word of the part can be confirmed, it is determined that there is a high possibility of having proofreading content suitability (S7). On the other hand, when the replacement word cannot be confirmed in the translation word candidate, it is determined that the possibility of having the proofreading content suitability is low (S6).

  For example, as an example that cannot be confirmed, it is assumed that the previous saun noun is “discussion”. The “explanation” Japanese translation candidate does not include “discussion”. Moreover, there are “argument, discussion, controversy, dispute, debate, issue, colloquy, disputation, haggle”, etc., but there is no “explanation” among them. Therefore, within the scope of the translation dictionary, “explanation” and “discussion” are considered to have no correspondence in both directions.

  In this way, if the replacement word is the first language, the dictionary is looked up from the first language using the translation dictionary of the second language. If the word is in the second language, the dictionary is looked up from the second language using the translation dictionary in the first language. Judge that it has.

  As a result, the locations where the calibration was performed were classified into “things that are unlikely to have proofreading content suitability” and “things that are likely to have proofreading suitability”, but “possibility to have proofreading content suitability” Since the “low-performing property” needs to be visually confirmed, it is possible to make the operator's convenience by devising the display such as highlighting the portion corresponding to this classification.

  Next, in FIG. 3, in determining the suitability of the proofreading content, if the proofreading classification is any of sentence movement, sentence combination, and sentence division, it is determined that the possibility of having the proofreading contents is high. If the proofreading classification is a combination or addition of deletion and addition, if the deleted or added word is a content word, it is determined that the necessity for checking the proofreading content is relatively high, and In some cases, it is possible to determine that the necessity of checking the calibration contents is relatively low.

  FIG. 19 is a flowchart showing an example of the case where the proofreading content suitability determination means 32 performs a proofreading content suitability determination process when the proofreading classification is a combination or addition of deletion and addition.

  The proofreading content aptitude determination means 32 reads the proofreading classification of the target bilingual document from the proofreading classification storage unit 31 (S1). In this case, the target bilingual document is the sentence of the first language document of the bilingual document shown in FIG. Then, it is determined whether or not the calibration classification is a combination of deletion and addition (S2).

  If the proofreading classification is not a deletion and addition combination or addition, the process ends. If the proofreading classification is a deletion and addition combination or addition, the proofreading content aptitude determination means 32 is targeted from the analysis document storage unit 26. The analysis information of the bilingual document is read to obtain attribute information such as part of speech (S3).

  Next, it is determined whether or not the part of speech of the content word is included in the part of speech of the overlapping part of the bilingual document before and after the proofreading (S4). (S5) If it is included, it is determined that the necessity of confirmation of the calibration content is relatively high (S6).

  As described above, when the proofreading classification is changed by a combination of deletion and addition, that is, when there is a replacement or only addition, the different parts of the parallel translation document before and after the proofreading are content words or function words Determine whether. This is because the content word serves to convey the meaning of the sentence and most of the content and is a word that has meaning alone, whereas the function word is a word that has almost no independent meaning other than showing grammatical relations. Yes, it is based on the idea that changing the content word is more likely to change the semantic content than changing the function word alone and requires more attention.

  By part of speech, nouns, adjectives, verbs, and adverbs belong to content words, and pronouns, prepositions (in English), conjunctions, and interjections belong to function words.

  FIG. 20 is an explanatory diagram of an example of a bilingual document when FIG. 19 is applied. In case 1 of FIG. 20, “of” is inserted after sequencing, and in case 2, human is replaced with “entire”. Here, “of” in case 1 is a preposition, “human” before proofreading in case 2 is a noun, and “entire” after proofreading is an adjective. Therefore, in the case of FIG. 20, since the function word is added in the case 1, the necessity for checking the proofreading contents is relatively low, and in the case 2 the replacement of the contents words, the necessity for checking the proofreading contents is relatively low. It will be expensive.

  In the above description, the case of English is assumed. However, in the Japanese classification, the classification of independent words and ancillary words is more common. The parts of speech that belong to independent words are nouns, pronouns, conjunctions, adverbs, conjunctions, emotional verbs, verbs, adjectives, and adjective verbs, and parts of speech that belong to an adjunct are auxiliary verbs and particles. Therefore, this can be applied to the above and the same processing can be performed. In addition, it is possible to improve the convenience of the operator by devising the display such as highlighting a portion that is determined to have a relatively high need for confirmation of calibration contents.

  Next, the determination of the suitability of the calibration content by the user will be described. FIG. 21 is a flowchart showing the processing contents of the proofreading content confirmation operation means 34 when the user determines the suitability of the proofreading contents, and FIG. 22 is an explanatory diagram of an example of a parallel translation document when FIG. 21 is applied.

  In FIG. 21, first, the control unit 24 activates the calibration content confirmation operation unit 34 according to a command from the input device 18 by the user. The proofreading content confirmation operation means 34 reads the bilingual document before and after proofreading from the bilingual document storage unit 26, and displays the different parts on the display device 17 so that they can be identified (S1).

  The difference between the sentence before the proofreading and the sentence after the proofreading is that, in the sentence of sentence number 1 in FIG. 22, the portion that was “owner or user” before the proofreading is “user or owner” after the proofreading. In addition, the user is notified that the part that has become “owner etc.” before the calibration has changed to “user etc.” after the calibration. This is notified by highlighting or color-coded display by the calibration content confirmation operation means 34.

  Next, the proofreading confirmation operation means 34 asks the user to approve the i-th proofreading result of the proofreading sentence (S2). This is performed, for example, by displaying information requiring approval on the display device 17. In addition, i is sequentially increased from 1 by 1.

  Then, the proofreading confirmation operation means 34 determines whether or not there is an approval from the user (S3), and if the approval is received from the user, the approval is requested whether the approved proofreading content may affect the entire document (S4). ), It is determined whether or not there is approval from the user (S5). If there is approval from the user, the proofreading confirmation operation means 34 confirms the approval, creates a matrix in which the changed parts before and after the calibration are extracted (S6), and the processing of the sentence of interest ends. FIG. 23 is an explanatory diagram of an example of a matrix for proofreading content approval.

  On the other hand, if the answer is not related to the whole document, that is, if the user answers that it is a local correction, the processing of the sentence of interest is terminated as it is, and the proofreading sentence is the last. Is determined (S7), 1 is added to i (S8), and the process returns to step S2.

  For example, when the user approves the change from “owner etc.” to “user etc.” of sentence number 1 in FIG. In the sentence, “owner etc.” is not changed to “user etc.”.

  If the user does not approve in step S3, the proofreading confirmation operation means 34 asks for approval to restore the calibration result (S9), and determines whether there is approval from the user (S9). S10). If the answer is to restore, the calibration content confirmation operation means 34 restores the calibration result (S11).

  For example, in FIG. 22, in the case where the option of returning to the original is taken in response to the change from “owner or user” to “user or owner”, the sentence is “the driver of the vehicle on the site, etc. The police officers, etc., when the instructions given in the same paragraph cannot be given to the driver, etc. due to lack of In other words, the method of parking the vehicle is immediately changed, or the vehicle should be moved from a place where the parking is prohibited, or the vehicle in the time-limited parking section is parked. From the fact that it should move from the place where it is ... ”,“ If there is no driver etc. of the vehicle at the site, the police officer etc. Is the owner or user of the vehicle Change the parking method of the vehicle immediately, or to move the vehicle from the place where the parking is prohibited, or for the time It changes to the effect that it should move from the place where the vehicle in the restricted parking section is parked.

  If it is determined in step S10 that there is no approval to restore, and after the restoration, the proofreading content confirmation operation means 34 inquires whether or not the current proofreading sentence has been corrected (S12). It is determined whether or not there is a correction request from the user (S13). If there is a correction request from the user, the correction request processing is performed (S14). If there is no correction request in step S13, and after the correction request processing is performed, it is determined whether or not the sentence to be corrected is the last (S7), 1 is added to i (S8), and the process returns to step S2. .

  In this way, if the user is instructed to confirm that the proofreading location of the proofreading sentence is valid, the proofreading content is retained, and if there is a correction request from the user, the correction is accepted and the proofreading content is corrected. To do.

  As described above, according to the embodiment of the present invention, the following effects can be expected. First, the conventional proofreading system checks the document, which was the result of machine translation, but has been expanded to include human translation results, expanding its use. In addition, the conventional proofreading system has dealt with the case where only one language has changed, but it also supports the case where both are changed in parallel in the actual situation. In addition, as a precondition, it is sufficient that the sentence structure is supported, and there are few restrictions.

  Secondly, to manually check that the content changes made on the source text are reflected on the translated text side without any excess or deficiency, and that the content changes made on the translated text side are reflected without any excess or deficiency, Although it uses a lot of nerves, this makes it possible to reduce leaks since the candidate areas for checking are identified. The user has read the whole sentence up to now, and for each sentence, the sentence before the first language proofreading, the sentence after the first language proofreading, the sentence before the second language proofreading, and the sentence after the second language proofreading. Although it was necessary to compare and review, and a great deal of labor was required, in the embodiment of the present invention, the time required for checking can be reduced, and the time can be used for creative work such as ingenuity of the expression itself. . Thereby, the document in the first language and the document in the second language can always be kept up-to-date.

DESCRIPTION OF SYMBOLS 11 ... Bilingual document proofreading apparatus, 12 ... Operation control apparatus, 13 ... Processor, 14 ... Memory, 15 ... Bilingual document proofreading program, 16 ... Work area, 17 ... Display apparatus, 18 ... Input device, 19 ... Mouse, 20 ... Keyboard , 21 ... disk drive, 22 ... hard disk drive, 23 ... input processing unit, 24 ... control unit, 25 ... storage device, 26 ... parallel translation document storage unit, 27 ... translation dictionary unit, 28 ... document analysis means, 29 ... analysis information Storage unit 30... Calibration content classifying unit 31... Calibration classification storage unit 32. Calibration content suitability judging unit 33... Output processing unit 34.

Claims (11)

A storage device storing a bilingual document proofreading program and a translation dictionary unit;
Enter the first language document before proofreading, the second language document before proofreading, the first language document after proofreading, and the second language document after proofreading for the first language document and the second language document that is the translation result. Together with an input device for inputting information necessary for operation,
A display device for displaying information necessary for proofreading a bilingual document;
In the bilingual document proofreading apparatus comprising an arithmetic control device that performs the bilingual document proofreading program,
The first language document before proofreading input from the input device is paired with the second language document before proofreading, and the first language document after proofreading and the second language document after proofreading are paired as a bilingual document. A bilingual document storage unit for storing;
Analyzing the first language document before proofreading, the second language document before proofreading, the first language document after proofreading, and the second language document after proofreading stored in the bilingual document storage unit using the translation dictionary unit Document analysis means for storing the analysis information in the analysis information storage unit,
Based on the analysis information stored in the analysis information storage unit, the first language document before proofreading and the first language document after proofreading, and the second language document before proofreading and the second language document after proofreading are compared. Calibration content classification means for classifying the calibration content and storing the calibration classification in the calibration classification storage unit;
Calibration content suitability judging means for judging the suitability of the calibration content based on the analysis information stored in the analysis information storage unit and the calibration classification stored in the calibration classification storage unit;
A bilingual document proofreading apparatus characterized by comprising:   The proofreading content classifying means, when there is no difference due to the proofreading between the first language document before proofreading and the first language document after proofreading, the first language document after proofreading and the second language document after proofreading The proofreading classification is the same, and when there are different parts due to the proofreading, they are classified into one of deletion, addition, sentence movement, sentence division, sentence combination, sentence deletion, and sentence addition. Bilingual document proofing device.   The calibration content suitability determination means displays and outputs the calibration content suitability result on the display device, prompts the user to confirm the calibration content suitability result displayed on the display device, and inputs it from the user via the input device. 3. The bilingual translation according to claim 1, further comprising proofreading content confirmation operation means for confirming the proofreading contents of the first language document after proofreading and the second language document after proofreading based on the operated instruction. Document proofing device.   4. The proofreading content confirmation operation means excludes proofreading contents determined to have proofreading content suitability by the proofreading content suitability judging means from the object of confirmation and correction work. The bilingual document proofing device.   The proofreading content aptitude determination means determines that the proofreading classification is not appropriate when the proofreading classification is deletion or addition, and determines that the proofreading content is appropriate when the proofreading classification is sentence movement, sentence deletion, or sentence addition. The bilingual document proofreading apparatus according to claim 2, wherein:   When the proofreading classification is deletion or addition, the proofreading content aptitude determination means determines the proofreading content aptitude when the part of speech of the deleted word matches the part of speech of the word added by replacement of the deleted word. 3. The bilingual document proofreading apparatus according to claim 2, wherein the bilingual document proofreading apparatus is judged to have.   When the proofreading classification suitability is deletion or addition, the proofreading content suitability judging means matches the part of speech of the deleted word with the part of speech of the word added by replacement of the deleted word, and each word 3. The bilingual document proofreading apparatus according to claim 2, wherein a translation dictionary portion of the corresponding language is looked up in a dictionary and it is determined that the proofreading content is appropriate when they are synonyms or different notations.   The proofreading content aptitude judgment means is that the proofreading classification is deletion or addition, the sentence structure of the first language document before proofreading and the sentence structure of the second language document after proofreading, or the sentence structure of the second language document before proofreading And the sentence structure of the first language document after proofreading have a correspondence in sentence pattern, and the part of speech of the deleted word matches the part of speech of the word added by replacing the deleted word. 3. The bilingual document proofreading apparatus according to claim 2, wherein the bilingual document proofreading apparatus is determined to be suitable.   The proofreading content aptitude judgment means is that the proofreading classification is deletion or addition, the sentence structure of the first language document before proofreading and the sentence structure of the second language document after proofreading, or the sentence structure of the second language document before proofreading And the sentence structure of the first language document after proofreading have a correspondence in sentence pattern, and the part of speech of the deleted word matches the part of speech of the word added by replacement of the deleted word If it is the first language, the dictionary is looked up using the translation dictionary of the first language from the first language, and if there is a corresponding second language word among the translation candidates, it is determined that the proofreading content is suitable, If the word is in the second language, the dictionary is looked up using the translation dictionary from the second language to the first language, and if there is a corresponding word in the first language in the translation candidate, it is judged that the proofreading content is suitable The bilingual document proofreading apparatus according to claim 2.   When the proofreading classification is a combination or addition of deletion and addition, if the word that has been deleted or added is a content word, the proofreading content aptitude determination means is relatively necessary to confirm the proofreading content. 2. The bilingual document proofreading apparatus according to claim 1, wherein the bilingual document proofreading apparatus determines that the necessity of checking the proofreading content is relatively low when it is determined to be high and is a function word.   The calibration content confirmation operation means leaves the calibration content when the user is instructed that the calibration location is valid, and accepts the correction content and corrects the calibration content when there is a correction request operation command from the user. The bilingual document proofreading apparatus according to claim 3.

Images