JP2017059151A - Bilingual dictionary creation device, bilingual dictionary creation method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a bilingual dictionary creation device and the like capable of creating a correct bilingual dictionary even for words whose frequency of appearance is small.SOLUTION: A bilingual dictionary creation method includes steps of: acquiring a corresponding sentence in which a Japanese sentence and an English sentence are associated with each other; determining a term pair between different languages out of the corresponding sentence; calculating a score which shows strength of correspondence with respect to the term pair, based on the appearance position of the term in Japanese related to the term pair appearing in the Japanese sentence and the appearance position of the term in English related to the term pair appearing in the English sentence; and creating the term pair as a bilingual dictionary between different languages according to the score and outputting it.SELECTED DRAWING: Figure 5

Description

  The present invention relates to a technique for creating a bilingual dictionary between different languages.

  Conventionally, it is known to create word pairs having the same semantic content between different languages. For example, in a known automatic dictionary creation method, correspondence data between a source language and a target language is read from a bilingual corpus, and based on the likelihood of the source language word and the target language word indicated in the correspondence data. There is one that associates these words (Patent Document 1).

JP 7-28819 A

  In the conventional dictionary automatic creation method, word pairs between different languages are created based on likelihood. However, for words that appear infrequently in the bilingual corpus, the likelihood (= probability) based on the appearance frequency is the same value or an equivalent value, so the likelihood necessary to create a correct word pair is obtained. In other words, there was a problem that an accurate bilingual dictionary could not be created.

  The present invention has been made in the above-described situation, and it is an object of the present invention to provide a bilingual dictionary creation device and the like that can create an accurate bilingual dictionary even for words with a low appearance frequency.

  The present invention for solving the above problems is a bilingual dictionary creation method in which a computer creates a bilingual dictionary, from a bilingual corpus in which a first language sentence and a second language sentence are associated in advance in sentence units, Obtaining a correspondence sentence; determining a term pair between different languages extracted from the correspondence sentence as a bilingual dictionary creation target; and a first language term relating to the term pair as the first language. Based on the appearance position appearing in the sentence and the appearance position where the second language character string related to the term pair appears in the second language sentence, the strength of the correspondence between the terms constituting the term pair is determined. And a step of creating and outputting the term pair as a bilingual dictionary between different languages according to an evaluation result by the evaluation unit.

  Here, the appearance positions of the character strings may be rearranged so that sentence structural features between the different languages are the same.

  The appearance position of the term is the order of appearance of the term in the corresponding language sentence, and the evaluating step includes a term that summarizes the morpheme or plural morphemes included in the corresponding language sentence, and the order of appearance of the term. Based on the relationship, a score representing the strength of the correspondence may be calculated, and the outputting step may create a bilingual dictionary between the different languages according to the score.

  The evaluating step may calculate all scores for the same term pair and determine a final score for the same term pair when the same term pair is obtained from different corresponding sentences. It may be.

  The present invention for solving the above-described problems includes an acquisition unit that acquires a corresponding sentence from a bilingual corpus in which a first language sentence and a second language sentence are associated in advance on a sentence basis; A determining unit that determines the extracted term pairs between different languages as a bilingual dictionary creation target, an appearance position where a term in the first language related to the term pair appears in the first language sentence, and a term related to the term pair Based on the appearance position at which the character string of the second language appears in the second language sentence, the evaluation unit that evaluates the strength of the correspondence between the terms constituting the term pair, and the evaluation result by the evaluation unit And an output unit that generates and outputs the term pair as a bilingual dictionary between different languages.

  According to the present invention, it is possible to create an accurate bilingual dictionary even for words with a low appearance frequency.

It is a figure which shows the example of a schematic structure of the whole bilingual dictionary creation system containing the bilingual dictionary creation apparatus of embodiment of this invention. It is a figure which shows the structural example on the hardware of the bilingual dictionary creation apparatus of FIG. It is a figure for demonstrating the outline of bilingual dictionary creation implement | achieved by the bilingual dictionary creation apparatus. It is a figure which shows an example of a function structure of a bilingual dictionary creation apparatus. It is a flowchart which shows an example of the whole process in a bilingual dictionary creation apparatus.

  Hereinafter, a schematic configuration of an entire bilingual dictionary creation system including a bilingual dictionary creation device according to an embodiment of the present invention will be described with reference to FIG. FIG. 1 is a diagram showing a schematic configuration example of the entire bilingual dictionary creation system 1.

  In FIG. 1, a bilingual dictionary creating system 1 includes a communication terminal 10, a bilingual dictionary creating device 30 that can be connected to the communication terminal 10 via a communication network 20 such as the Internet, and an external that can be connected to the bilingual dictionary creating device 30. The system includes a parallel corpus 40 as a system.

  In the bilingual dictionary creation system 1, HTTP (HyerText Transfer Protocol) communication is performed between the communication terminal 10 and the bilingual dictionary creation device 30, but other communication methods may be used.

  The communication terminal 10 includes a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), a display device such as a liquid crystal display, and an input device such as a touch panel. In this embodiment, the communication terminal 10 is a laptop computer as an example, but may be a portable terminal, a PDA (Personal Digital Assist), a personal computer, or the like.

  The bilingual corpus 40 is a document database that is provided on, for example, a network transmission path and can communicate with the bilingual dictionary creation device 30 on the network. As will be described later, the bilingual corpus 40 stores sentences having the same semantic content between different languages (such as English and Japanese) in association with each other. The bilingual corpus 40 includes a central processing unit (CPU), a read only memory (ROM), and a random access memory (RAM).

[Hardware configuration of bilingual dictionary creation device]
Next, a hardware configuration example of the bilingual dictionary creation device 30 shown in FIG. 1 will be described with reference to FIG. FIG. 2 is a diagram illustrating a configuration example of the bilingual dictionary creation device 30.

  As shown in FIG. 2, the bilingual dictionary creation apparatus 30 includes a CPU (Central Processing Unit) 31, a ROM (Read Only Memory) 32, a RAM (Random Access Memory) 33, a communication interface 34, and an external access unit 35. Is a server device.

  The CPU 31 is connected to each component via a bus and transfers control signals and data, and executes a program for realizing the entire processing of the bilingual dictionary creation device 30 and performs arithmetic processing.

  The ROM 32 stores a bilingual dictionary creating program necessary for the operation of the entire bilingual dictionary creating device 30. The bilingual dictionary creating device 30 of the present embodiment is realized by executing the program. Of course, it is possible to take in necessary programs and data using a cloud service instead of the ROM 33.

  The above-described program may be stored in a storage medium such as a CD-ROM.

  The RAM 33 temporarily stores a program for performing a bilingual dictionary creation process, which will be described later, and various data.

  The communication interface 34 has a network interface function and performs communication with the communication terminal 10.

  The external access unit 35 is an interface for the CPU 31 to access the parallel corpus 40 and communicate with the parallel corpus 40. In this embodiment, a corresponding sentence to be described later in the parallel corpus 40 is transmitted to the CPU 31 via the external access unit 35. The bilingual dictionary creation device 30 may be configured as a single server device that also has a function of reading parallel translations of different language sentences in the bilingual corpus 40.

[Outline of bilingual dictionary creation]
Next, an outline of bilingual dictionary creation realized by the bilingual dictionary creating apparatus 30 will be described with reference to FIGS. FIG. 3 is a diagram for explaining the outline of bilingual dictionary creation. (A) shows different language sentences A and B associated in advance, (b) morpheme analysis processing, and (c) terminology An extraction process, (d) a term pair determination process, (e) a term rearrangement process, (f) a term pair evaluation process, and (g) a bilingual dictionary creation process are shown. 3A to 3G merely illustrate the bilingual dictionary creation process.

  First, in this bilingual dictionary creation device 30, at the time of bilingual dictionary creation processing, corresponding sentences (parallel translation sentences) in which different language sentences are associated in advance in units of sentences are read from the bilingual corpus 40. In the example of FIG. 3A, the CPU 31 of the bilingual dictionary creation device 30 performs the Japanese sentence A “He works for a company in Tokyo” and the English sentence “He works for a company in Tokyo”. A corresponding sentence including B is read.

  Next, as illustrated in FIG. 3B, the CPU 31 performs, for example, morphological analysis as a process of dividing each of the Japanese sentence A and the English sentence B into morphemes of the smallest unit that has linguistic meaning. Then, the CPU 31 uses “he” as a term in each sentence A and B (in this embodiment, for example, an independent word whose meaning can be expressed by the term itself), for example, as shown in FIG. , “Tokyo”, “Company”, “Work”, “he”, “works”, “company” and “tokyo” are extracted, and for example, as shown in FIG. Create a pair of terms (such as a pair of “he” and “he”).

  As shown in FIG. 3E, the CPU 31 “he works for a company in Tokyo” so that sentence structural features (grammatical structure, meaning of terms, etc.) are the same between Japanese and English. The English sentence B term "" is rearranged and converted into a rearranged sentence B1 "Hega Tokyo in company for works". In other words, the rearranged sentence B1 is obtained by subjecting the English sentence B to postfix conversion so that it matches the word order of the Japanese sentence A. Note that if the sentence structural features between different languages are the same or similar, the CPU 31 may not perform the rearrangement process.

  Then, as shown in FIG. 3F, the CPU 31 determines the appearance position of “company” in the Japanese sentence A (the appearance from the beginning of the sentence is the seventh word) and the appearance position of “company” in the rearrangement sentence B1. (Appearance from the beginning of the sentence is the fifth word), the strength of the correspondence between the two terms constituting the term pair of “company” and “company” is evaluated. In the following description, the “fifth word” and “seventh word” indicating the number of morphemes that appear from the beginning of the sentence are referred to as “order of appearance”.

  In this embodiment, as an example of evaluating the strength of the correspondence, {(order of appearance of terms included in Japanese A) / (total number of morphemes included in Japanese sentence A) − (included in rearrangement sentence B1) The absolute value of the value given by the expression (1) of the order of appearance of terms / (total number of morphemes contained in the rearrangement sentence B1)} is obtained as a score. In the example of FIG. 3F, the appearance order of “company” in Japanese A = seventh word; the total number of morphemes contained in Japanese A = 12 words; the appearance order of “company” in rearrangement sentence B1 = 5 Word: Since the total number of morphemes contained in the rearranged sentence B1 is 8, the score is about 0.04 from {(7/12)-(5/8)}.

  According to the example of FIG. 3G, the CPU 31 also obtains scores of other term pairs, for example, term pairs ((he, he), (Tokyo, tokyo), (company) having a score of 0.04. , Company)) as a bilingual dictionary. That is, it can be evaluated that the smaller the score is, the stronger the correspondence between the terms constituting the term pair is. Therefore, the correct bilingual dictionary can be created by considering the score value.

[Functional structure of bilingual dictionary creation device]
Next, the functional configuration of the bilingual dictionary creation device 30 will be described with reference to FIG. FIG. 4 is a diagram illustrating an example of a functional configuration of the bilingual dictionary creation device 30 realized on the hardware configuration illustrated in FIG. 2.

  In FIG. 4, the bilingual dictionary creation device 30 includes an acquisition unit 301, a determination unit 302, an evaluation unit 303, and an output unit 304. The determination unit 302 includes a term extraction unit 3021 and a term pair creation unit 3022. These components are appropriately referred to in the following description of the processing of the bilingual dictionary creation device 30.

[Processing of bilingual dictionary creation device]
Hereinafter, the process of the bilingual dictionary creation device 30 executed to realize this bilingual dictionary creation will be described with reference to FIGS. FIG. 5 is a flowchart showing an example of the entire processing in the bilingual dictionary creation device 30.

  In FIG. 5, CPU31 acquires the correspondence sentence between different languages from the bilingual corpus 40 (step S10). As shown in the example in Fig. 3 (a), the correspondence sentences are Japanese sentence A "He works for a company in Tokyo" and English sentence B "He works for a company in Tokyo". is there.

  In step S <b> 10, the CPU 31 functions as the acquisition unit 301 in cooperation with the external access unit 35.

  Next, the CPU 31 determines a term pair extracted from the correspondence sentences between different languages acquired in step S10 as a bilingual dictionary creation target (step S11). In FIG. 3 (d), a term pair such as (he, he) is determined and the term pair is created. Before the determination, the CPU 31 performs processing shown in FIG. 3 (b) and FIG. As shown in (c), for example, Japanese sentence A and English sentence B are each subjected to morphological analysis, and independent words such as “he” and “he” are extracted as terms.

  In the determination process of step S <b> 11, the CPU 31 functions as the determination unit 302. In the term extraction process, the CPU 31 functions as the term extraction unit 3021. In the term pair creation process, the CPU 31 functions as the term pair creation unit 3022.

  CPU31 calculates the score of a corresponding term pair for all the term pairs determined in step S11. In this case, the CPU 31 rearranges the English sentence B in order to match Japanese sentence structure characteristics (grammatical structure, semantic content) before calculating the score. In FIG. 3 (e), for example, English sentence B is subject to postfix conversion and set as rearrangement sentence B1, and in this rearrangement sentence B1, the appearance order of “company” is the fifth from the beginning of the sentence (FIG. 3). (F)). After this rearrangement is performed, the CPU 31 is based on the appearance order of the “company” in the Japanese sentence A (= the seventh word) and the appearance order of “company” in the rearrangement sentence B1 (= the fifth word). Then, the score of the term pair “company” and “company” is calculated. The score at this time is, as shown in the above formula (1), {(order of appearance of “company” included in Japanese A) / (total number of morphemes included in Japanese sentence A) − (sort sentence B1 {(7/12) − (5/8)} = about 0.04 from the relational expression “order of appearance of“ company ”included in” / (total number of morphemes included in rearrangement sentence B1)}.

  The smaller the score described above, the stronger the correspondence between the terms constituting the term pair, so it is possible to evaluate whether the correspondence between the term pairs is strong or not based on the score.

  In step S <b> 12, the CPU 31 functions as the evaluation unit 303.

  In FIG. 5, the CPU 31 sequentially and repeatedly executes the processes of steps S <b> 11 and S <b> 12 for all corresponding sentences acquired from the bilingual corpus 40 in step S <b> 10.

  The target may not be all the corresponding sentences but only a part of the specified corresponding sentences.

  The rearrangement process is performed in step S12, but may be performed before the score is calculated (step S10 or step S11).

  The CPU 31 creates and outputs a term pair as a bilingual dictionary based on the score calculated in step S12 (step S13). For example, when the score is equal to or higher than a preset threshold, a term pair is created and output as a bilingual dictionary. In FIG. 3G, for example, since the threshold is set at 0.04, a term pair having a score of 0.04 ((he, he), (Tokyo, tokyo), (company, company)) Is created and output as a bilingual dictionary. The bilingual dictionary may be created as long as the term pair is shown as a bilingual dictionary, and can be implemented by various methods such as a list or a dictionary format. The output destination is, for example, the communication terminal 10.

  In step S <b> 13, the CPU 31 functions as the output unit 304 in cooperation with the communication interface 34.

  As described above, according to the bilingual dictionary creation device 30 of the present embodiment, by calculating the score for a term pair based on the appearance position of each term pair in the corresponding sentence of different language sentences, A bilingual dictionary between different languages is created. Here, since the score is calculated based on the appearance position, unlike the appearance frequency, a different value is obtained depending on the appearance position even for a term having a low appearance frequency. This makes it possible to create an accurate bilingual dictionary even for terms that appear less frequently.

  The above embodiment may be changed.

  For example, in the score calculation process of FIG. 5 (step S12), when the same term pair is acquired from different corresponding sentences, all scores for the same term pair are calculated, and the final term for the same term pair is calculated. A reasonable score. As the final score, for example, an arithmetic mean, a geometric mean, or the like is applied.

  In the above-described score calculation, a weight may be given according to the appearance frequency of the term in the term pair.

  In the above, the processing for calculating the score using (the total number of morphemes contained in the language sentence) has been described with reference to the above formula (1), but instead of this, ( The total number) may be used. For example, when a sentence composed of a plurality of continuous morphemes becomes a noun after dividing a language sentence by morphemes, the CPU 31 recognizes a group of morphemes composed of a plurality of morphemes as one term and calculates the score be able to.

  The bilingual dictionary creation process described above may be applied to languages other than Japanese and English as long as the bilingual dictionary is created in consideration of the score based on the appearance position.

DESCRIPTION OF SYMBOLS 10 Communication terminal 30 Bilingual dictionary creation apparatus 40 Bilingual corpus 301 Bilingual acquisition part 302 Determination part 303 Evaluation part 305 Output part 3021 Term extraction part 3022 Term pair creation part

Claims (6)

A bilingual dictionary creation method in which a computer creates a bilingual dictionary,
Obtaining a corresponding sentence from a bilingual corpus in which the first language sentence and the second language sentence are associated in advance in sentence units;
Determining a pair of terms between different languages extracted from the corresponding sentence as a translation target;
Based on the appearance position where the term of the first language related to the term pair appears in the first language sentence, and the appearance position where the character string of the second language related to the term pair appears in the second language sentence, Evaluating the strength of the correspondence between terms that make up the term pair;
Creating a bilingual dictionary between different languages and outputting the term pair according to an evaluation result by the evaluating unit.   The bilingual dictionary creation method according to claim 1, wherein the appearance positions of the character strings are rearranged so that sentence structural features between the different languages are the same. The appearance position of the term is the appearance order of the term in the corresponding language sentence,
The evaluating step calculates a score representing the strength of the correspondence relationship based on a relationship between a morpheme or a morpheme included in the corresponding language sentence and an appearance order of the terms,
3. The bilingual dictionary creating method according to claim 1, wherein the outputting step creates a bilingual dictionary between the different languages according to the score.   In the evaluation step, when the same term pair is obtained from different corresponding sentences, all the scores for the same term pair are calculated to determine a final score for the same term pair. The method of creating a bilingual dictionary according to claim 3. An acquisition unit for acquiring a correspondence sentence from a parallel corpus in which a first language sentence and a second language sentence are associated in advance in sentence units;
A determination unit for determining a term pair between different languages extracted from the corresponding sentence as a bilingual dictionary creation target;
Based on the appearance position where the term of the first language related to the term pair appears in the first language sentence, and the appearance position where the character string of the second language related to the term pair appears in the second language sentence, An evaluation unit that evaluates the strength of the correspondence between terms constituting the term pair;
A bilingual dictionary creation device, comprising: an output unit that creates and outputs the term pairs as bilingual dictionaries between different languages according to the evaluation result by the evaluation unit.   A bilingual dictionary creating program for causing a computer to execute the bilingual dictionary creating method according to any one of claims 1 to 4.

Images