JP2001331484A - Recording medium recording a parallel thesaurus generation program, recording medium recording a parallel thesaurus, and recording medium recording a parallel thesaurus navigation program - Google Patents

Abstract

(57) [Summary] [Problem] To automatically generate a parallel thesaurus composed of concept associations in a language and concept connection between languages. Realizes text mining using the generated parallel thesaurus. A server computer 1 includes a Japanese corpus 51.
, And a Japanese thesaurus generation 10 that generates a Japanese term-related thesaurus 71 by analyzing the correlation between the terms. English is extracted by extracting the terms from the English corpus 52 and analyzing the correlation between the terms. English thesaurus generation 2 to generate term-related thesaurus 72
0, a Japanese-English thesaurus combining 30 that combines a Japanese term-related thesaurus 71 and an English term-related thesaurus 72
And

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for generating a parallel thesaurus by combining a thesaurus of two languages from a text corpus of two languages, and a program for a navigation system using the parallel thesaurus.
And a recording medium on which a parallel thesaurus is recorded.

[0002]

2. Description of the Related Art With the increase of electronic text information, the importance of information access technology is increasing. The present inventors have previously applied for a thesaurus navigation method as Japanese Patent Application No. 11-28101. This method
This is a technique to efficiently extract useful information (hereinafter referred to as text mining) from a document database (hereinafter referred to as a text corpus) that stores text data. Is extracted to generate a thesaurus, and the contents of the thesaurus are displayed on the browser of the client terminal for navigation.

[0003] In addition, IPSJ Database System Research Group / Informatics Basic Research Group Research Report DBS-118-1
A text mining system based on the above navigation method is reported in 3 / FI-54-13 "Related thesaurus navigation for corpus" (May 17, 1999).

[0004]

In the field of information retrieval,
There is a growing need to search foreign language documents by inputting a search request expressed in a native language. This information search method is called cross-language information search and is being actively studied.

[0005] A typical method of cross language information retrieval is, for example, IPSJ Transactions Vol. 40, No. 11, page 4
075-4086, "A Study on English-Japanese / Japanese-English Cross-Language Search Using Machine Translation" (November 1999). This search method uses a bilingual dictionary or a machine translation system to translate a search request into the same language as a document and then execute a document search. In this case, since the search request is shorter than the document and has less context information, there is a problem that it is difficult to translate with high accuracy and the search accuracy is low.

To solve this problem, the above-mentioned Japanese Patent Application No. 11-28 is disclosed.
It is conceivable to use the navigation method proposed in No. 101. In this case, it is considered that the above problem in the cross-language information search can be solved by extending from a monolingual language to a bilingual language and using it as a front end of the information search system. Need to be done.

On the other hand, the technique of generating a thesaurus from a text corpus is effective not only for the above-described application to text mining but also for various natural language processing application systems, but has the following technical problems. I have.

[0008] The conventional technology for automatically generating a thesaurus has a problem with respect to the handling of polysemous words. In the related art, the relation between terms is extracted.However, since the relation between terms is inherently semantic, it is necessary to divide a polysemy term into meanings, that is, concepts, and to extract the relation between concepts. Is ideal. According to the prior art, for example, "loan", "money", "river" and "wate" are related terms of "bank".
r ”etc. are all extracted regardless of the concept of“ bank ”.

Here, "loan" and "money" are terms related to "bank" as an institution for depositing and withdrawing money, and include "river" and "water".
Is a related term of "bank" as a waterside location. Therefore, we divide "bank" into the concepts it represents, and associate related terms such as "loan" and "money" with "rive
It is desirable that relevant terms such as "r" and "water" be extracted separately.

[0010] The conventional thesaurus automatic generation technology is as follows.
There is also a problem with the handling of synonyms. In the related art, a related term is extracted based on a co-occurrence probability, that is, a probability of appearing in the vicinity of the text. For this reason,
Synonyms are not even extracted as related terms and are treated as separate entities. Since synonyms represent the same concept, it is desirable that synonyms are handled collectively by one entity.

Accordingly, an object of the present invention is to solve the above-mentioned problems in the conventional technology. The first purpose is to create a parallel thesaurus consisting of concepts and relationships between concepts in the first language, concepts and concepts in the second language, and connections between concepts in the first language and concepts in the second language. An object of the present invention is to provide a recording medium recording a parallel thesaurus generation program which is automatically generated from a text corpus of one language and a second language.

A second object is to record a parallel thesaurus navigation program as a front-end for cross-language information retrieval, which enables high-precision translation of retrieval requests, particularly focusing on handling of polysemy and synonyms. It is another object of the present invention to provide a recording medium that has been designed. The third purpose is
An object of the present invention is to provide a recording medium on which a parallel thesaurus is recorded in order to realize effective text mining using the parallel thesaurus.

[0013]

SUMMARY OF THE INVENTION The present invention provides a computer for extracting terms from a text corpus of a first language,
First language thesaurus generation means for generating a first language term-related thesaurus by analyzing the correlation between terms; and extracting terms from a text corpus of the second language;
Second language thesaurus generation means for generating a second language term-related thesaurus by analyzing the correlation between terms; a thesaurus coupling means for coupling the first language term-related thesaurus with the second language term-related thesaurus; And a computer-readable recording medium on which a program for causing the apparatus to function as a parallel thesaurus generation device including the program is recorded.

[0014] Further, the Cirrus combining means is combined with Term combining means for combining corresponding terms between the term related thesaurus in the first language and the term related thesaurus in the second language with reference to the bilingual dictionary. First language concept label generating means for generating a concept label from each term in the first language by combining terms in the second language, and a concept from each term in the second language by combining the combined terms in the first language A second language concept label generating means for generating a label, a concept connecting means for converting a term connection between languages into a concept connection between languages, and a relation between terms included in a term relation thesaurus of the first language between concepts. A first language concept relation thesaurus generating means for converting the relation into terms, and a relation between terms included in the term relation thesaurus in the second language into relations between concepts. And a second language concept related thesaurus generating unit that merges concepts of the first language that are combined with the same concept of the second language and have a similar set of related concepts of the first language into one concept A first language concept merging unit, and a second language concept that is merged into one concept by merging concepts of the second language that are combined with the same concept of the first language and have a similar set of related concepts of the second language. Conceptual merging means. This makes it possible to generate a parallel thesaurus that takes into account the concepts of polysynonyms and synonyms.

The parallel thesaurus generator operates as follows. The first language thesaurus generating means includes:
By extracting a set of terms and co-occurring terms from the text corpus of the language and analyzing the correlation between the terms,
A set of related terms is output for each term. When the first language is Japanese, for example, as a set of related terms of the term “bank”, {loan, interest, account, interest, securities, economy,
Finance, investment} is output.

The second language thesaurus generating means executes the same processing as the first language thesaurus generating means on the text corpus of the second language, and outputs a set of related terms for each term. When the second language is English, for example,
The set of related terms for the term "bank" is {account, r
iver, interest, loan, boat, investment, fishing, park, ec
onomy, lake} is output.

In the thesaurus coupling means, each means operates as follows. The term combining unit refers to the bilingual dictionary and combines corresponding terms between the term-related thesaurus in the first language and the term-related thesaurus in the second language. For example, "bank" and "bank" are combined,
And "bank" are combined.

The first language concept label generating means generates at least one concept label from each term in the first language by combining the terms in the second language combined. Similarly, the second language concept label generating means generates at least one concept label from each term in the second language by combining the terms in the first language combined. For example, "bank" is joined to "bank" and "shore". At this time, the related term set of "bank" and "shore"
If the related term sets are not similar, "bank" and "shore"
Are conceptually different from each other, and two concept labels “bank / bank” and “bank / shore” are generated from “bank”.

The concept linking means converts the term link between languages into a concept link between languages. For example, the term combination "bank-bank" is converted to the concept combination "bank-bank-bank", and the term combination "shore-bank" is converted to the concept combination "bank-bank."
Is converted to "shore".

The first language concept related thesaurus generating means includes:
The relation between terms included in the term relation thesaurus of the first language is converted into the relation between concepts. Similarly, the second language concept relation thesaurus generation means converts the relation between terms included in the term relation thesaurus of the second language into the relation between concepts. For example, the relation between terms “bank-interest” is converted into the relation between concepts “bank / bank-interest / interest / interest”, and the relation between terms “bank-river” is between the concepts “bank / bank-interest”. river ".

The first language concept merging means uses the same second language concept merging means.
A first language concept that is combined with a language concept and has a similar set of related first language concepts is merged into one concept. Similarly, the second language concept merging unit merges concepts of the second language, which are combined with the same concept of the first language and have a similar set of related concepts of the second language, into one concept. For example, two Japanese concepts, "interest rate" and "interest rate"
Are combined with the English concept of interest / interest. At this time, if the set of related concepts of “interest” and the set of related concepts of “interest” are similar, “interest” and “interest” are merged into one concept “interest−interest”.

As described above, each function operates to construct a concept relation thesaurus of the first language composed of concepts of the first language and the relation between the concepts, and a concept relation concept of the second language and the relation between the concepts of the second language. Then, a parallel thesaurus is generated in which the concept related thesaurus of the second language is combined.

Also, the present invention provides a method for combining a term in a first language with a term in a second language.
This is a recording medium that records a parallel thesaurus composed of a concept label combining language terms and a combination of concept labels based on the concept represented by the concept label.

Further, the present invention causes a computer to function as a parallel thesaurus navigation system including transition means for transitioning from a set of a plurality of terms or concepts in a first language to a set of a plurality of terms or concepts in a second language. Computer-readable recording medium on which a program for recording is recorded.

[0025] The transition means may convert a set of terms or concepts of the first language into a term or concept of the second language connected to the terms or concepts in the set, and convert the terms or concepts of the second language. A transition is made to a set in which terms or concepts that are related terms or concepts and are not combined with the terms or concepts in the first language are added. In a parallel thesaurus navigation system, the means for transitioning from the set of concepts in the first language to the set of concepts in the second language operates as follows.

The concept of the second language combined with the concept in the concept set before the transition, and the concept of the second language that is strongly related to the concept of the second language and is not combined with the concept of the first language Are combined to form a concept set after the transition. As a result, it is possible to make a transition from the concept set of the first language to the related concept set of the second language, including concepts that are not explicitly connected in the parallel thesaurus.

[0027]

Embodiments of the present invention will be described below in detail with reference to the accompanying drawings. FIG. 1 is a block diagram illustrating a configuration of a parallel thesaurus generation device according to an embodiment of the present invention and a parallel thesaurus navigation system including the device. In the present embodiment, a description will be given of the generation of a parallel thesaurus in Japanese and English as a parallel thesaurus in which a thesaurus of two languages is combined.

The parallel thesaurus navigation system (hereinafter, referred to as a system) according to the present embodiment includes a server computer 1 and a client computer 2 connected to each other via a communication network 3.

The server computer 1 performs a parallel thesaurus generation process for associating a Japanese thesaurus with an English thesaurus, and performs a thesaurus search process among the processes of the present system. The server computer 1 includes a processing device 11,
It is mainly composed of the input device 12 and the storage device 13.

The processing device 11 executes the entire processing of the server computer 1. In particular, it executes each data processing related to the generation of the parallel thesaurus shown in FIGS. The input device 12 is a CD-ROM drive, a floppy (registered trademark) disk drive, or the like, depending on the text corpus acquisition medium, and is used to input the text corpus.

The storage device 13 is a general term for storage means such as a RAM, a ROM, and a magneto-optical disk library device (not shown). For example, a processing program of the server computer 1 is fixedly stored in the ROM. Data and work files created in the course of processing of the server computer 1 are stored in a RAM.
The corpus, thesaurus and bilingual dictionary (see FIG. 2) of each language are stored in a large-capacity storage device such as a magneto-optical library device.

The client computer 2 performs the display of the thesaurus transmitted as the search result of the server computer 1 and the interactive processing with the user among the processing of the present system. Next, the details of the parallel thesaurus generation processing will be described with reference to FIGS.

FIG. 2 is a diagram functionally explaining input / output data and a module configuration in the parallel thesaurus generation device. The input of the parallel thesaurus generator is a Japanese-English bilingual text corpus in which a Japanese corpus 51 and an English corpus 52 are paired. The Japanese corpus 51 and the English corpus 52 are subject to the condition that they are texts in the same field, but need not be a translation.

The output of the parallel thesaurus generator is a Japanese concept related thesaurus 61, an English concept related thesaurus 62, Japanese-English combined data 63, English-Japanese combined data 6
4 Japanese-English parallel thesaurus. The combined data 63 and the combined data 64 have the same information but different record formats. Although both are redundant, both are output in consideration of the efficiency of the Japanese-English thesaurus combining process.

The modules constituting the parallel thesaurus generation device are a Japanese thesaurus generation 10, an English thesaurus generation 20, and a Japanese-English thesaurus combination 30. The Japanese thesaurus generation 10 generates a Japanese term-related thesaurus 71 from the Japanese corpus 51. The English thesaurus generator 20 generates an English term-related thesaurus 72 from the English corpus 52. The Japanese-English thesaurus combination 30
Referring to the Japanese-English bilingual dictionary 73 and the English-Japanese bilingual dictionary 74, the Japanese term-related thesaurus 71 and the English term-related thesaurus 72 are used to convert the Japanese concept-related thesaurus 61, the English concept-related thesaurus 62, the Japanese-English concept combined data 63, And the English-Japanese concept combined data 64 is generated. The Japanese-English bilingual dictionary 73 and the English-Japanese bilingual dictionary 74 have the same information content but different record formats. Although redundant, both are used in consideration of the efficiency of the Japanese-English thesaurus combining process.

FIG. 3 is a diagram for explaining the details of the processing of the Japanese thesaurus generation 10. As shown in FIG. 3, the process of generating a Japanese thesaurus includes three steps: term extraction 101, co-occurrence data extraction 102, and correlation analysis 103.

(1) Term Extraction 101 Terms are extracted from the Japanese corpus 51, and the appearance frequency is counted. As terms, nouns and compound nouns whose appearance frequency is equal to or greater than a predetermined threshold are extracted. Compound nouns are extracted by pattern matching using a part-of-speech sequence pattern. Among the high-frequency words, there are many general words that are not particularly related to the field. They are removed using a stopword list.

With respect to this stop word list, for example, by using “above” as the stop word of the head element, a noun phrase such as “above system” can be excluded. Similarly, a noun phrase such as "entire system" can be excluded by setting "entire" as a stop word of the last element.

(2) Co-occurrence Data Extraction 102 A pair of co-occurring terms is extracted, and the co-occurrence frequency is counted. Window co-occurrence is adopted as the definition of co-occurrence. That is, while moving a window having a certain width along the text, a pair of terms included in the window at each position is extracted. The width of the window is, for example, 25 terms excluding the function word. (3) Correlation analysis 103 A statistical correlation value is calculated for all the term pairs, and a term pair having a correlation value equal to or greater than a predetermined threshold is extracted. Mutual information is used as a term correlation value.

As a result of the above steps (1) to (3), a Japanese term-related thesaurus 71 is obtained. The Japanese term-related thesaurus 71 is a set of records representing a related term set for each Japanese term. That is, RT _J (x _i ) = {x (i, 1), x (i, 2),..., X (i, m _i )} (i =
1, 2,..., M). Here, x _i is a Japanese term, x (i, m _i ) is a related term,
i is number assigned to each term, m _i is related terms the number of the i Japanese term, M is the total number of Japanese terms. An example of a record is shown below.

RT _J (bank) = {loan, interest rate, account, interest rate, securities, economy, finance, investment}. RT _J (interest rate) = {loan, lending, savings, raising, bank}. The processing of the Japanese thesaurus generation 10 has been described above.
The processing of the English thesaurus generation 20 is exactly the same as the Japanese thesaurus generation 10. As a result of the English thesaurus generation processing, an English term-related thesaurus 72 is obtained. The English term related thesaurus 72 is a set of records representing a related term set for each English term. That is, RT _E (y _i ) = {y (i, 1), y (i, 2),..., Y (i, n _i )} (i =
1, 2,..., N). Where y _i is the English term, y (i, n _i ) is the related term, i
The number assigned to each term, n _i is related terms the number of the i English term, N is the is the total number of English terms. An example of a record is shown below. _{RT E (bank) = {account} , river, interest, loan, boat, inv
estment, fishing, park, economy, lake}. RT _E (interest) = {loan, deposit, bank, sciene, econom
y, exchange, politics}.

FIG. 4 shows a Japanese-English thesaurus combining module 3
FIG. 7 is a diagram for explaining the details of the process of 0. As shown in FIG. 4, the Japanese-English thesaurus combining process includes a Japanese-English term combining 30
1. Japanese concept label generation 302, English concept label generation 303, Japanese-English concept combination 304, Japanese concept related thesaurus generation 305, English concept related thesaurus generation 306, Japanese concept merge 307, and English concept merge 308
Consists of two steps. Hereinafter, details of these processes will be described.

(1) Japanese-English Term Combination 301 Referring to the English-Japanese bilingual dictionary 73 and the English-Japanese bilingual dictionary 74, the corresponding terms between the Japanese term-related thesaurus 71 and the English term-related thesaurus 72 are combined, and The English term combination data 91 and the English-Japanese term combination data 92 are output.
Since the Japanese term-related thesaurus 71 and the English term-related thesaurus 72 have already been described in the input of the Japanese-English term combination, the Japanese-English bilingual dictionary 73 and the English-Japanese bilingual dictionary 74 will be described.

The Japanese-English bilingual dictionary 73 is a collection of records representing a bilingual English term set for each Japanese term. _{That, D JE (a i) =} {b (i, 1), b (i, 2), ···, b (i, l i)} (i = 1,
2, ..., K). Here, a is a Japanese term and b is an English term. An example of a record is shown below. D _JE (bank) = {bank}. D _JE (shore) = {bank}.

The English-Japanese bilingual dictionary 74 is a set of records representing a bilingual Japanese term set for each English term. _{That, D EJ (b i) =} {a (i, 1), a (i, 2), ···, a (i, k i)} (i =
1, 2,..., L). Here, b is an English term and a is a Japanese term. An example of a record is shown below. D _EJ (bank) = {bank, bank, bank}. D _EJ (interest) = {interest, interest, interest}.

Next, the output of the Japanese-English term combination will be described. The English-Japanese term combination data 91 and the English-Japanese term combination data 92 represent the same information in different formats. Although both are redundant, both are output in consideration of the efficiency of the subsequent processing. The Japanese-English term combination data 91 is a set of records representing a set of English terms combined with each of the Japanese terms. That is, TL _JE (x _i ) = {y ′ (i, 1), y ′ (i, 2),..., Y ′ (i, n ′ _i )}
(i = 1, 2,..., M). Here, x is a Japanese term and y 'is an English term.

The English-Japanese term combination data 92 is a set of records representing a set of Japanese terms combined with each English term. That is, TL _EJ (y _i ) = {x ′ (i, 1), x ′ (i, 2),..., X ′ (i, m ′ _i )}
(i = 1, 2,..., N). Here, y is an English term and x 'is a Japanese term.

The algorithm of the Japanese-English term combination 301 is as follows. 1) Initialize the Japanese-English term combination data 91. That is, TL _JE (x _i ) ← φ (i = 1, 2,..., M). 2) Initialize the English-Japanese term combination data 92. That is, TL _EJ (y _i ) ← φ (i = 1, 2,..., N).

3) A Japanese term x and an English term y satisfying the following two conditions are combined. (A) The bilingual relation <x, y> is supported by the bilingual dictionary. (B) Domain relevance DR (x, y) of bilingual relationship <x, y>
Is greater than or equal to a predetermined threshold. That is, for all pairs of Japanese terms x and English terms y that satisfy (a) and (b), TL _JE (x) ← TL _JE (x) ∪ {y} and TL _EJ (y) ← TL _EJ
(y) {{x}. Execute (A) x is a sequence of _k terms x ₁ , x ₂ ,..., X _k , y is k
Number of term y _1, y _2, ···, a sequence of _{y k, {y '1,} y'
_{2, ···, y 'k}} = {y 1, y 2, ···, y as is y _{k}' 1} (∈
D _JE (x ₁ )), y ' ₂ (∈D _JE (x ₂ )), ..., y' _k (∈D _JE (x _k ))
Exists. (B) DR (x, y) ≧ θ.

According to the condition (a), in order to know whether the Japanese term x and the English term y are in a bilingual relation, it is determined whether or not a bilingual relation is established between the constituent elements by determining the order of the terms in the set. Checked regardless. In particular, when k = 1, y∈D _JE (x). That is, it means that the term pair is registered in the bilingual dictionary. When k ≧ 2, it means that the bilingual relationship between the constituent elements is a pair of compound terms registered in the bilingual dictionary. Based on the condition (b), it is checked whether the bilingual relationship suggested by the bilingual dictionary is a relationship that holds in the domain. Translation <
The domain relevance DR (x, y) of x, y> is defined by the following equation.

[0051]

(Equation 1)

DR _JE (x, y) is a ratio of Japanese terms related terms whose English translation is English terms among the related terms. That is, the appearance context indicating the context in which a certain Japanese term appears is the degree to which the appearance context of the English term overlaps. DR _EJ (x, y)
Is the degree to which the appearance context of the English term overlaps the appearance context of the Japanese term. If the appearance contexts have some degree of commonality, it can be considered that a bilingual relationship is established in the domain, so the threshold value θ of the domain relevance should be set to a small value. By the above algorithm, the Japanese-English term combination 301
Is executed.

(2) Japanese concept label generation 302 (3) English concept label generation 303 Japanese concept label generation 302 and English concept label generation 30
The processing of 3 is exactly the same except that the roles of Japanese and English are reversed. Therefore, here, the English concept label generation 303 will be described.

The English concept label generation step 303 generates one or more English concept labels from each of the English terms based on the English-Japanese term combination data 92 and the Japanese term-related thesaurus 71. Further, a set of related terms is generated for each of the generated English concept labels. For this purpose, the English term related thesaurus 72, the Japanese-English term combined data 91, the Japanese term related thesaurus 71 and the English-Japanese term combined data 92 are referred to.

Since the input data of the English concept label generation 303 has already been described, the output data will be described. An English concept label is a combination of terms and is defined as follows. <English concept label>: = <English term> | <English term> / <Japanese modifier> | <English concept label> + <English concept label>. <Japanese modifier>: = <Japanese term> | <Japanese modifier> / <Japanese term>.

<English terms> / <Japanese terms> /.../
<Japanese term> indicates a concept common to the concept represented by the Japanese term among the concepts represented by the English term. For example, "bank / bank" indicates a bank as an organization that performs business related to money, and "bank / bank" indicates a bank as a location along a river or lake. <English concept label> + <
The English concept label> designates, as a core, a common part of the concept indicated by the two concept labels, and a concept in a range in which the concept indicated by each concept label is combined. `` Duty / tax + ta
x "," plane / airplane + airplane ", and" reasoning + in
ference "is an example.

The English concept label data 94 is a record C representing a set of English concept labels corresponding to each English term y.
A collection of _E (y). 4 shows an example of a set of English concept labels. C _E (interest) = {interest / interest / interest / interest rate}. This record C _E (interest) contains TL _EJ (interest) = {interest, interest, interest rate} in the English-Japanese term combination data 92. When a record is included, it is generated correspondingly.

The English concept related term data 96 is a set of records representing a related term set for each English concept label, and is described as follows. RT _E (Y _i ) = {y (N + i, 1), y (N + i, 2),..., Y (N + i, n _{N + i} )}
(i = 1, 2,..., Q). Here, Y is an English concept label, and y is an English term.

The English concept related term data 96 is intermediate data for generating the English concept related thesaurus 62 as the final purpose. What is required as the English concept relation thesaurus 62 is not a relation term set but a relation concept set. However, a related concept set cannot be created until a concept label set for all terms has been generated. Therefore, a related term set is tentatively created, and is converted into a related concept set in the subsequent English concept related thesaurus generation 306.

A set of English concept labels for the English term y
C_E(y) and C_ERelated terms for concept labels in (y)
The algorithm for generating the set is as follows. 1) The English term y is connected to at least one Japanese term
I) Create initial data of English concept label set. English
Modify each of the Japanese terms combined with term y to Japanese
Generate an English concept label as a child and use it as its element. You
That is, C_E(y) ← {y · x | x∈TL_EJ(y)}. ii) If the similarity between two English concepts is greater than or equal to a predetermined threshold α
If so, allow them to be integrated into one English concept
Repeat as much as possible. That is, While ∃y · x₁/ x_Two/… / X_k(∈C_E(y)) and y ・ x '₁/ x '_Two/.../X '_{k '}(∈C_E(y)) s.t. S (y ・ x₁/ x_Two/.../X_k, Y ・ x '₁/ x '_Two/.../X '_{k '}) ≧ α, C_E(y) ← C_E(y)-{yx₁/ x_Two/.../X_k, Y ・ x '₁/ x '_Two/.../X '_{k '}} + {Yx₁/ x_Two/.../X_k/ x '₁/ x '_Two/.../X '_{k '}}. Here, two English concepts Y related to a common English term y
₁= Yx₁/ x_Two/.../X _kAnd Y_Two= Yx '₁/ x '_Two/.../X '_{k '}
Similarity S (Y₁, Y_Two) Is defined by the following equation.

[0061]

(Equation 2)

That is, it is defined by the degree of overlap between related term sets of Japanese modifiers constituting a concept label. iii) Create related term set data for each of the English concepts obtained as a result of process ii). English concept y
The related term set RT _{E of} x ₁ / x ₂ /.../x _k (y · x ₁ / x ₂ / ···
* / X _k ) is as follows.

[0063]

(Equation 3)

Here, JM1 is a set of elements of the Japanese modifier. _{That, JM1 = {x 1, x} 2, ···, x k}. J
M2 is a set of Japanese terms combined with the English term y, with the Japanese qualifier element removed. That is,
_{JM2 = TL EJ (y) -JM1} . 2) When the English term y is not combined with the Japanese term x i) The English term y itself is used as a concept label. The English concept label set has this as the only element. That is, C _E (y) ← {y}. ii) The related term set RT _E (y) of the English term y is used as it is as the related term set of the English concept label y. The setting of the threshold value α in 1) ii) of the above algorithm will be supplemented. The purpose here is to obtain a Japanese modifier to distinguish between multiple concepts represented by one English term. Therefore, the threshold value α is set smaller,
It is better to combine synonymous Japanese translations into one Japanese modifier.

The Japanese concept label generation 302 is the same as the English concept label generation 303. The Japanese concept label data 93, which is the output, is a set of records C _J (x) representing a Japanese concept label set corresponding to each Japanese term x, similarly to the English concept label data 94. The Japanese concept label is similar to the English concept label and is a combination of terms as follows.

<Japanese concept label>: = <Japanese term> | <Japanese term> · <English modifier> | <Japanese concept label> + <Japanese concept label>. <English modifier>: = <English term> | <English modifier> / <
English terms>. The related term data 95 of the Japanese concept, which is another output of the Japanese concept label generation 302, is the same as the related term data 96 of the English concept, and is a set of records representing a related term set for each Japanese concept label. is there. That is, RT _J (X _i ) = {x (M + i, 1), x (M + i, 2),..., X (M + i, n _{M + i} )}
(i = 1, 2,..., P). Here, X is a Japanese concept label, and x is a Japanese term.

(4) Japanese-English Concept Combination 304 The Japanese-English concept combination 304 receives the Japanese concept label data 93 and the English concept label data 94, and converts the Japanese-English concept combination data 63 and the English-Japanese concept combination data 64. Generate.
Japanese concept label data 93 and English concept label data 9
4 has already been explained.
3 and the English-Japanese concept combination data 64 will be described.

The Japanese-English combined data 63 is a set of records representing a set of English concepts combined with each Japanese concept. That is, CL _JE (X _i ) = {Y ′ (i, 1), Y ′ (i, 2),..., Y ′ (i, q ′ _i )}
(i = 1, 2,..., P). Here, X is a Japanese concept label, and Y 'is an English concept label.

Similarly, the English-Japanese concept combined data 64 is a set of records representing a set of Japanese concepts combined with each English concept. That is, CL _EJ (Y _i ) = {X ′ (i, 1), X ′ (i, 2),..., X ′ (i, p ′ _i )}
(i = 1, 2,..., Q). Here, Y is an English concept label, and X 'is a Japanese concept label.

The algorithm for generating the combined Japanese-English concept data 63 and the combined English-Japanese concept data 64 is as follows. Relative to 1) all Japanese concept labels _{X = x · y 1 / y} 2 / ··· / y k ', CL JE (X) = {Y | Y = y · x 1 / x 2 / ·· _{· / x k (∈C E (} y)), y∈ {y 1, y 2, ···, y k '}, {x 1, x 2, ···, x k} ∋x}. That is, a set of elements Y of the English concept set C _E (y) of the English term y included in the English qualifier of X and including x in the Japanese qualifier is generated.

2) All English concept labels Y = y · x ₁ / x ₂
/.../X _{k '} , CL _EJ (Y) = {X | X = x · y ₁ / y ₂ /.../y _k ({C _J (x)), x} {x _{1, x 2, ···, x} k '}, {y 1, y 2, ···, y k} ∋y}. That is, the Japanese term x included in the Japanese modifier of Y
, A set of elements X of the Japanese concept set C _J (x) that includes y in the English modifier.

An output example of the Japanese-English concept combination 304 is shown. The Japanese concept label data 93 is C _J (interest) = {interest / interest}, C _J (interest) = {interest / interest}, C _J (interest) = {interest / interest}, and the English concept label data 94 Is C _E (interest) = {interest / interest / interest / interest / interest}. At this time, CL _JE (interest / interest) = {interest / interest}, CL _JE (interest / interest) = {interest / interest / interest}, CL _JE (interest / interest) = {interest・ Interest / interest} is generated, and CL _EJ (interest / interest) = {interest / interest}, CL _EJ (interest / interest / interest) = {interest / interest, interest / in
terest} is generated.

Here, one rule regarding the notation of the concept label is determined. If there is only one English concept corresponding to an English term, it is meaningless to attach a Japanese qualifier, and the term itself can be used as a concept label. The same applies to Japanese terms. Unlike the algorithm of the Japanese-English concept combination 304 described above, the subsequent processing does not include the judgment based on the Japanese modifier or the English modifier. Therefore, for the concept that is the only concept of the term, the concept label is changed to the term itself when the concept combination data 63 and 64 are output. According to this rule, the concept data 63
Is CL _JE (interest) = {interest / interest}, CL _JE (interest) = {interest / interest / interest}, CL _JE (interest) = {interest / interest / interest}, and the English-Japanese concept combined data 64 is CL _EJ (interest / interest) = {interest}, CL _EJ (interest / interest / interest) = {interest, interest}. Here, since "interest" is a word representing a single concept, the concept label "interest / interest" is abbreviated to "interest". The same applies to "interest rate / interest" and "interest rate / interest", which are abbreviated as "interest rate" and "interest rate", respectively.
On the other hand, "interest" is a word that represents multiple concepts,
Concept labels "interest / interest""interest / interest / interest rate"
Cannot be abbreviated.

(5) Japanese Concept Related Thesaurus Generation 30
5 (6) English concept related thesaurus generation 306 The Japanese concept related thesaurus generation 305 inputs the Japanese concept label data 93 and the Japanese term related term data 95 and outputs the Japanese concept related thesaurus 61. The English concept related thesaurus generation 306 receives the English concept label data 94 and the English concept related term data 96 and outputs the English concept related thesaurus 62.
These inputs have been described.

Output Japanese Concept Related Thesaurus 61
And the English concept related thesaurus 62 are as follows. The Japanese concept related thesaurus 61 is a set of records representing a related concept set of each Japanese concept. That is, RC _J (X _i ) = {X (i, 1), X (i, 2),..., X (i, p _i )} (i =
1,2, ..., P). Here, X is a Japanese concept label. An example of a related concept set is shown below.

RC _J (bank) = {loan, interest, account, interest, securities, economy, finance, investment}. RC _J (shore) = {river, water, boat, lake, fishing}. The English concept related thesaurus 62 is a collection of records representing a related concept set of each English concept. That is, RC _E (Y _i ) = {Y (i, 1), Y (i, 2),..., Y (i, q _i )} (i =
1,2, ..., Q). Here, Y is an English concept label. An example of a related concept set is shown below. RC _E (bank) = {account, account, interest, interest / rate, loan, investment, economy}. RC _E (bank / shore) = {river, boat, water, fishing, park / park, lake}.

The algorithm of the English concept related thesaurus generation 306 is as follows. The algorithm of the Japanese concept related thesaurus generation 305 is exactly the same. Each element y of the related term set RT _E (Y) of the English concept Y
, The element of the concept label set _CE (y) having the highest degree of correlation with Y is selected as an element of the related concept set RC _E (Y). That is,

[0078]

(Equation 4)

Here, S ₂ is the degree of correlation of the English concept based on the related term set, and is defined by the following equation. S ₂ (Y ₁ , Y ₂ ) = | RT _E (Y ₁ ) ∩RT _E (Y ₂ ) | / | RT
_E (Y ₁ ) ∪RT _E (Y ₂ ) |. For example, the related term set of the English concept “bank / bank” includes “interest”, and the concept label set of the English term “interest” is {interest / interest, interest Interest rate / interest rate}. At this time, "bank / bank" and "interest / interest"
Correlation with "bank / bank" and "interest / interest / interest rate"
Is calculated. If the latter has a high degree of correlation,
As an element of the related concept set of "bank
"Interest / interest" is selected.

(7) Japanese Concept Merge 307 (8) English Concept Merge 308 The processing of the Japanese concept merge 307 and the English concept merge 308 are exactly the same except that the roles of Japanese and English are reversed. Therefore, here, the English concept merge 308 will be described.

The English concept merge 308 merges English concepts combined with the same Japanese concept having a high degree of similarity into one concept. The input is an English concept related thesaurus 62, Japanese-English concept combined data 63, English-Japanese concept combined data 6
4 and the output is their updated data.

The algorithm of the English concept merge 308 is as follows. The following processing is repeated as much as possible for all Japanese concepts X. If _{Y 1, Y 2 ∈CL JE (} X) by _{S 3 (Y 1, Y 2} ) ≧ the set Y ₁ of β becomes English concepts, Y ₂ is present, executes c) from a). Here, S ₃ (Y ₁ , Y
₂ ) is the similarity between the English concepts Y ₁ and Y ₂ and is defined by the following equation. S ₃ (Y ₁ , Y ₂ ) = | RC _E (Y ₁ ) ∩RC _E (Y ₂ ) | / | RC
_E (Y ₁ ) ∪RC _E (Y ₂ ) |.

A) Update of English Concept Related Thesaurus 62 For all Y∈RC _E (Y ₁ ), RC _E (Y) ← RC _E (Y)
− {Y ₁ } + {Y ₁ + Y ₂ }. For all Y∈RC _E (Y ₂ ), RC _E (Y) ← RC _E (Y)
− {Y ₂ } + {Y ₁ + Y ₂ }. RC _E (Y ₁ + Y ₂ ) ← RC _E (Y ₁ ) ∪RC _E (Y ₂ ). Erase RC _E (Y ₁ ) and RC _E (y ₂ ).

B) Update of Japanese-English concept combined data 63 All x６３CL_EJ(Y₁), CL_JE(X) ← CL
_JE(X)-{Y₁} + {Y₁+ Y _Two}. All X∈CL_EJ(Y_Two), CL_JE(X) ← CL
_JE(X)-{Y_Two} + {Y₁+ Y _Two}. c) Update of English-Japanese concept combined data 64 CL_EJ(Y₁+ Y_Two) ← CL_EJ(Y₁) ∪CL_EJ(Y_Two). CL_EJ(Y₁) And CL_EJ(Y_Two).

It is preferable that the threshold value β in the above algorithm is set to be relatively large, and only concepts having a very high similarity are merged. This is because using a term with a different concept range or nuance as a separate entity results in a more useful thesaurus. This is different from the case where the purpose is to distinguish a plurality of concepts represented by the terms of the partner language (the threshold α in the algorithm of the English concept label generation 303).

By the above-described processing, the Japanese-English bilingual text corpus in which the Japanese corpus 51 and the English corpus 52 are paired, the Japanese concept-related thesaurus 61, the English concept-related thesaurus 62, the Japanese-English concept combined data 63,
A Japanese-English parallel thesaurus composed of English-Japanese concept combined data 64 can be generated. The Japanese-English parallel thesaurus generated in this manner is connected to the communication network 3 shown in FIG.
Is used for thesaurus navigation by the client computer 2 via the. Next, the processing of this system will be described with reference to FIGS.

FIG. 5 is a diagram for explaining the contents of the display screen of the client computer 2 in the present system. FIG.
Is composed of a concept grouping area 1010, a zoom-in area 1020, and a function selection button.
The function selection buttons include a zoom-in button 1030, a translation button 1040, a clear button 1050, an end button 1
060.

In the zoom-in area 1020, one or more concept clusters 1021 are displayed together with the selection buttons 1022 associated therewith. This concept cluster 1
021 is a set of highly relevant concepts. For example,
If it is a concept cluster corresponding to “global environmental issues”,
Concepts such as “global warming”, “ozone layer”, “greenhouse effect”, “CFC”, and “atmosphere” are displayed. The user of the client computer 2 can designate a plurality of these concept clusters 1021.

FIG. 6 is a flowchart for explaining the processing of the parallel thesaurus navigation system according to the present embodiment. Hereinafter, the processing of the present system will be described corresponding to the display contents shown in FIG. First, an initial screen is displayed (step 410). On the initial screen, the concept set area 1010 and the zoom-in area 1020 are blank.
The system includes a language indicator for switching between "Japanese" and "English". When the initial screen is displayed, the language indicator is set to "Japanese".

After displaying the initial screen (step 410), the system is in a state of waiting for input (step 420). In this state, concept set area 1010 is writable, and the user typically writes one or more Japanese terms or Japanese concept labels. Depending on the button pressed while waiting for input, branching occurs as follows.

(1) When the zoom-in button 1030 is pressed The concept set displayed in the concept set area 1010 is read (step 430). User sets concept collection area 1
Writing to 010 is usually a term, not a concept label. If a term has been written, processing is performed assuming that all concept labels generated from the term have been written. This process is performed by referring to the Japanese concept related thesaurus 61 when the language indicator is "Japanese" and referring to the English concept related thesaurus 62 when the language indicator is "English".

Next, the concept set is expanded by adding a concept strongly related to the concept included in the concept set, and the expanded concept set is clustered (step 440). This process is performed by referring to the Japanese concept related thesaurus 61 when the language indicator is "Japanese" and referring to the English concept related thesaurus 62 when the language indicator is "English". Finally, the obtained concept cluster is displayed together with the select button 1022 in the zoom-in area 10.
20 (step 450), and returns to the state of waiting for input.

(2) Concept cluster selection button 1022
Is pressed. The selected concept cluster 1021 is added to the concept set area 10
10 (overwrite) (step 460), and return to the state of waiting for input. In the state of waiting for input, the concept set area 1010 is writable, and the user can add or delete terms or concept labels.

(3) When translation button 1040 is pressed The concept set displayed in concept set area 1010 is read (step 470). This processing is performed in step 430
Is exactly the same as Next, the concept set is translated (step 480). When the language indicator is "Japanese", the translation from the Japanese concept set to the English concept set is executed, and when the language indicator is "English", the translation from the English concept set to the Japanese concept set is executed.

Finally, the translation result is stored in the concept set area 101.
0 is displayed (overwritten), the language indicator is inverted (step 490), and the process returns to the state of waiting for input. In the state of waiting for input, the concept set area 1010 is writable, and the user can add or delete terms or concept labels.

(4) When the clear button 1050 is pressed The display returns to the initial screen display state (step 410). (5) When the end button 1060 is pressed The processing ends. The above-described processing enables navigation of a parallel thesaurus including transition between languages.

FIG. 7 is a diagram for explaining in detail the concept set translation (step 480) which characterizes the present system. FIG. 7 shows a process of translating a Japanese concept set into an English concept set. The process of translating an English concept set into a Japanese concept set is exactly the same.

When the input Japanese concept set is given, the English concepts connected to the Japanese concepts in the Japanese concept set are collected with reference to the Japanese-English concept combination data 63 to form the core English concept set. Is generated (step 481).

Next, with reference to the English concept relation thesaurus 62, English concepts that are strongly related to the English concepts included in this English concept set are collected, and further, the English-Japanese concept combined data 64
And select a related English concept that is not combined with the Japanese concept. Add the selected English concept to the core English concept set to obtain the translation result (step 4
82).

An example of translation (transition) from a Japanese concept set to an English doubt set will be described. The Japanese concept set of input is {global warming, ozone layer, greenhouse effect, chlorofluorocarbon, atmosphere, carbon dioxide, environment}. It is assumed that the Japanese-English concept combined data 63 includes the following records.

CL _JE (global warming) = φ. CL _JE (ozone layer) = {ozone layer}. CL _JE (greenhouse effect) = φ. CL _JE (Freon) = φ. CL _JE (atmosphere) = {atmosphere / atmosphere}. CL _JE (carbon dioxide) = {carbon dioxide}. CL _JE (environment) = {environment}. Further, it is assumed that the English concept related thesaurus 62 includes the following records.

RC _J (ozone layer) = {chrolofluorocarbo
n, depletion, atmosphere, atmosphere, warming}. RC _J (atmosphere ・ atmosphere) = {pollution, environment, gas
・ Gas / gas, carbon dioxide}. RC _J (carbon dioxide) = {atmosphere, atmosphere, energy, war
ming, environment, regulation}. RC _J (environment) = {protection, carbon dioxide, ene
rgy, atmosphere ・ atmosphere, pollution}. The English-Japanese concept combined data 64 includes the following records.

CL _EJ (ozone layer) = {ozone layer}. CL _EJ (chrolofluorocarbon) = φ. CL _EJ (depletion) = {destruction}. CL _EJ (atmosphere, atmosphere) = {atmosphere}. CL _EJ (warming) = φ. CL _EJ (pollution) = {contamination}. CL _EJ (environment) = {environment}. CL _EJ (gas, gas / gas) = {gas, gas}. CL _EJ (carbon dioxide) = {carbon dioxide}. CL _EJ (energy) = {energy}. CL _EJ (regulation) = {regulation}. CL _EJ (protection) = {protection}.

At this time, the translation result from the Japanese concept set {global warming, ozone layer, greenhouse effect, Freon, atmosphere, carbon dioxide, environment} to the English concept set is {ozone layer, atmosphere, atmosphere, carbon dioxide, environment, chrolofluorocarbon, w
arming}.

Of the six English concepts that make up the English concept set, “ozone layer”, “atmosphere / atmosphere”, and “carb”
The four, "on dioxide" and "environment," were explicitly combined in the thesaurus with the Japanese concepts in the Japanese concept set. Also, two of “chrolofluorocarbon” and “warming” were not explicitly connected to the Japanese concept in the Japanese concept set in the thesaurus,
It has been added as a related concept of two English concepts.
In fact, "chrolofluorocarbon" is an English translation of "CFCs" and "warming" is part of the English translation of "global warming." In this way, it is possible to obtain a translation result including a bilingual translation that is not explicitly expressed as a concept combination.

The embodiments of the present invention have been described in detail with reference to the drawings. However, the specific configuration is not limited to these embodiments, and the design of the present invention does not depart from the gist of the present invention. May be changed. First, in the above-described embodiment, a parallel thesaurus of Japanese and English is generated. For example, a parallel thesaurus in general bilingual languages including Japanese and French, and Japanese is also used. It may be generated.

Further, in the above embodiment, as shown in FIG. 4, a function taking into account the concept of each of the polysemy and the synonym is realized. Alternatively, a function considering only synonyms may be realized.
This can be realized by selectively providing the functions of the Japanese concept merge 307 and the English concept merge 308.

The client computer 2 according to the present invention may be a personal computer or a workstation connected to the communication network 3 by a wired line, or a mobile communication terminal (mobile phone) connected to the communication network 3 by a wireless line. , PHS (Personal Han
dy-Phone System), PDA (Personal Digital Assist)
ance) etc.).

Note that the parallel thesaurus generation device and the parallel thesaurus navigation system of the present invention are also realized by a program for causing the server computer 1 or the client computer 2 to function. This program is stored in a computer-readable recording medium such as a CD-ROM.

The storage device 13 itself shown in FIG. 1 may be used as a storage device for storing a program for operating the parallel thesaurus generation device or the parallel thesaurus navigation system, or a CD-ROM as an external storage device. A program reading device (not shown) such as a ROM drive may be provided, and a CD-ROM or the like readable by inserting the program reading device may be used. The recording medium is a magnetic tape, cassette tape, floppy disk, hard disk, MO / MD / DVD, etc.
Alternatively, it may be a semiconductor memory.

The parallel thesaurus generated by the parallel thesaurus generating apparatus of the present invention is a CD-R.
It may be stored in a computer-readable recording medium such as OM. This parallel thesaurus is a combination of the thesauruses of two languages, and a concept label combining the terms of the two languages is generated by Japanese and English concept label generations 302 and 303.
04 (see FIG. 4), concept labels of two languages are combined based on the concept.

[0112]

According to the recording medium storing the parallel thesaurus generation program of the present invention, a parallel thesaurus in which related thesauruses of two languages are combined can be automatically generated from a text corpus of two languages. . The generated parallel thesaurus shows the relation between concepts.

Further, according to the recording medium on which the parallel thesaurus of the present invention is recorded, unlike the conventional thesaurus showing the relation between terms, the problem of polysemy and synonym is solved. By using the thesaurus obtained, the accuracy of various natural language processing systems can be improved.

According to the recording medium in which the parallel thesaurus navigation program of the present invention is recorded, efficient text mining over a plurality of languages can be performed. In particular, it becomes easier to navigate the thesaurus in the native language and access information in foreign languages. The translation accuracy of a search request, which is a problem in the conventional cross-language information search, is also greatly improved by the concept set transition (translation) function.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a parallel thesaurus generation device according to an embodiment of the present invention and a parallel thesaurus navigation system accommodating the device.

FIG. 2 is a diagram functionally illustrating input / output and module configuration in a parallel thesaurus generation device according to an embodiment of the present invention.

FIG. 3 is a diagram illustrating details of processing for generating a Japanese thesaurus.

FIG. 4 is a diagram illustrating details of the processing of the Japanese-English thesaurus combining module.

FIG. 5 is a diagram illustrating contents of a display screen of a client computer in the parallel thesaurus navigation system according to the embodiment of the present invention.

FIG. 6 is a flowchart illustrating a process of the parallel thesaurus navigation system according to the embodiment of the present invention.

FIG. 7 is a diagram illustrating in detail a concept set translation process (step 480) characterizing the parallel thesaurus navigation system according to the embodiment of the present invention.

[Explanation of symbols]

 Reference Signs List 1 server computer 2 client computer 3 communication network 10 Japanese thesaurus generation 11 processing device 12 input device 13 storage device 20 English thesaurus generation 30 Japanese-English thesaurus combination 51 Japanese corpus 52 English corpus 61 Japanese concept related thesaurus 62 English concept related thesaurus 63 Japanese-English concept combined data 64 English-Japanese concept combined data 71 Japanese term-related thesaurus 72 English term-related thesaurus 73 Japanese-English bilingual dictionary 74 English-Japanese bilingual dictionary 81 Terms and appearance frequency 82 Co-occurrence term pairs and co-occurrence frequency 91 days English term combination data 92 English-Japanese term combination data 93 Japanese concept label data 94 English concept label data 95 Japanese concept related term data 96 English concept related term data 101 Term extraction 102 Co-occurrence data extraction 103 Correlation analysis 301 Japanese-English term combination 302 Japanese concept label generation 303 English concept label generation 304 Japanese-English concept combination 305 Japanese concept related thesaurus generation 306 English concept related thesaurus generation 307 Japanese concept merge 308 English concept merge 1010 Concept set area 1020 Zoom-in area 1021 Concept cluster 1022 Select button 1030 Zoom-in button 1040 Translate button 1050 Clear button 1060 Exit button

Claims (5)

[Claims] 1. A first language thesaurus generating means for extracting a term from a text corpus of a first language and analyzing a correlation between the terms to generate a term-related thesaurus of the first language, and a second language Extracting the terms from the text corpus of the second language and analyzing the correlation between the terms to generate a term-related thesaurus of the second language, a term-related thesaurus of the first language and a term-related thesaurus of the second language A computer-readable recording medium on which is recorded a program for functioning as a parallel thesaurus generating device, comprising: a thesaurus coupling means for coupling a thesaurus. 2. The method according to claim 1, wherein the step of combining the plurality of terms includes a step of combining the corresponding terms between the term-related thesaurus in the first language and the term-related thesaurus in the second language with reference to the bilingual dictionary. First language concept label generating means for generating a concept label from each term in the first language by combining terms in the second language, and a concept from each term in the second language by combining the combined terms in the first language A second language concept label generating means for generating a label, a concept connecting means for converting a term connection between languages into a concept connection between languages, and a relation between terms included in a term relation thesaurus of the first language between concepts. A first language concept relation thesaurus generating means for converting to a relation, and a relation between terms included in a term relation thesaurus of a second language to a relation between concepts A second language concept related thesaurus generating means, and a first language concept associated with the same second language concept
First language concept merging means for merging concepts of a first language having a similar set of language concepts into one concept, and a concept of a second language associated with the same concept of the first language and related 2. The computer-readable recording medium according to claim 1, further comprising: a second language concept merging unit that merges concepts of a second language having a similar set into a single concept. 3. A parallel thesaurus characterized by recording a thesaurus composed of a concept label obtained by combining a term in a first language and a term in a second language and a combination of concept labels based on the concept represented by the concept label. The recording medium on which it was recorded. 4. A program for causing a computer to function as a parallel thesaurus navigation system including transition means for transitioning from a set of a plurality of terms or concepts in a first language to a set of a plurality of terms or concepts in a second language. A computer-readable recording medium characterized by being recorded. 5. The method according to claim 1, wherein the transition unit is configured to convert the term or concept of the second language from a set of terms or concepts of the first language to a term or concept of the second language connected to the terms or concepts in the set. 5. The computer-readable recording medium according to claim 4, wherein a transition is made to a set obtained by adding a term or concept related terms or concepts that are not combined with the terms or concepts in the first language.