JP2003099429A - Glossary generation device, glossary generation program and glossary retrieval device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a glossary generation device and glossary generation program for extracting a term and commentary data defining the term based on an attributive modification clause from natural language text data and to provide a term retrieving device for retrieving the commentary data of the word from the word. SOLUTION: The glossary generation device 1 includes a modification relation analytic means 11 for generating modification relation information of the clause of the text data, a glossary data extraction means 12 for extracting term data, a concept data extraction means 13 for extracting concept data showing the high-order concept of term data, a learning database 16 for registering learning data which becomes feature when the attributive modification clause becomes an explanation sentence defining the term, a modification data extraction means 15 for extracting the attributive modification clause defining term data as modification data, and an explanation data generating means 17 for generation the commentary data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a glossary generation device and a glossary generation program for extracting a term and data explaining the term from text data, and a glossary search for retrieving the explanation data of the term from the term. Regarding the device.

[0002]

2. Description of the Related Art Conventionally, from natural language text data,
As a method of extracting a term and commentary data defining the term, a method based on matching of surface layer patterns that represent surface characteristics of a sentence is known.

In this method, for example, when the term is α and the definition sentence of the term is β, “α is β” and “α
By matching the text data based on the surface pattern such as ".beta.", The term matched with the surface pattern and the commentary data defining the term can be extracted.

[0004]

However, when the news manuscripts used in actual broadcast programs are investigated, the surface patterns such as "α is β" and "α is β" are the news manuscripts. Only 7.7% of the total is used, to generate a glossary that extracts the terms and commentary data for the terms contained in the news manuscript,
There was a problem that the amount of data was insufficient.

In addition, to define a term, the adnominal modifier related to the term may be used. When using this adnominal modifier, for example, in a news manuscript, such as "realization of an electronic government that enables procedures such as obtaining a resident's card and permitting and conducting business by companies without going to the government office ..." In the adnominal qualification section, "enable procedures such as obtaining a resident's card or permitting or permitting a company to perform on the Internet without going to the government office," defines the term "e-government."

As described above, when the term is defined by the adnominal modification clause, there is a problem that the explanation data of the term cannot be extracted by the surface layer pattern matching in the above-mentioned conventional technique.

The present invention has been made in view of the above technical problems, and from natural language text data,
An object is to provide a glossary generation device and a glossary generation program that extract a term and commentary data defining the term, and a glossary search device that searches commentary data of the term from the term based on the adnominal modifier. And

[0008]

The present invention was devised in order to achieve the above-mentioned object. First, the terminology generation device according to claim 1 has the following configuration. That is, by performing a morphological analysis and a syntactic analysis on the input natural language text data, a dependency analysis means for generating dependency information of a clause of the text data, and a noun or a noun phrase from the text data. Based on the term data extracting unit that extracts a character string as term data, the dependency information, and a specific paraphrase expression that paraphrases the term data, conceptual data indicating a superordinate concept of the term data is extracted from the text data. Based on the dependency information and the learning data, the concept data extracting means, a learning database in which learning data that is a feature when the adnominal modifier is an explanatory text defining a term is registered in advance, and the term data based on the dependency information and the learning data. It is judged whether the adnominal modifier related to is the definition of the term data, and the adnominal modifier judged to be the definition And modifying the data extraction means for extracting as modified data, the based on the concept data and the modified data, and configured to and a description data generating means for generating a commentary data defining the term data.

According to this configuration, the glossary generating device is
By the morphological analysis and the syntactic analysis of the input natural language text data by the dependency analysis means,
Dependency information of the clause of this text data is generated, and by this term data extraction means, from this text data,
A character string that is a noun or a noun phrase is extracted as term data. The character string that serves as the noun or noun phrase is extracted by, for example, syntactic analysis. Then, based on the dependency information and a specific paraphrase expression that paraphrases the term data, the concept data extracting unit extracts concept data indicating a superordinate concept for the extracted term data from the text data.

Further, the terminology generation device uses the modification data extraction means to learn the modification information and a learning database in which learning data, which is a characteristic when the adnominal modification clause serves as an explanatory sentence defining a term, is registered in advance. Based on the data, the adnominal modifier that defines the term data is extracted as modification data, and the comment data generation means generates comment data defining the term data based on the conceptual data and the modification data. To do.

Thus, the terminology generation device performs the dependency analysis on the input natural language text data by morphological analysis or syntactic analysis to indicate the term data and the superordinate concept of the term data. Concept data and modification data defining the term data are extracted from the text data, and commentary data defining the term data is generated based on the concept data and the modification data.

Further, the terminology generation device according to claim 2 is the terminology generation device according to claim 1, further comprising a concept database for registering term data and a plurality of concept data corresponding to the term data. The concept data extraction means determines one concept data corresponding to the term data from the plurality of concept data based on the appearance frequency thereof.

According to this structure, the glossary generating device is
The concept data extracting means refers to the appearance frequency of the concept data in the concept database in which the term data and a plurality of concept data corresponding to the term data are registered, and determines one concept data corresponding to the term data. Thereby, the terminology generation device determines the concept data having the highest appearance frequency among the plurality of concept data corresponding to the term data in the concept database as the concept data most suitable for the term data.

The terminology generation device according to claim 3 is the terminology generation device according to claim 1 or 2, wherein the learning database is an explanatory sentence in which the adnominal modifier clause defines a term. The verb directly related to the term and the particle immediately preceding it are registered as learning data.

According to this structure, the terminology generation device is
The learning database registers, as learning data, the verb directly related to the term and the particle immediately preceding the term when the adnominal modifier is an explanatory text defining the term. As a result, the glossary generation device specifies, from a plurality of adnominal modifiers, an adjunct modifier that is a descriptive sentence that defines the term, by a combination of the verb directly related to the term and the particle immediately preceding it.

Further, the terminology generation device according to claim 4 is the terminology generation device according to any one of claims 1 to 3, wherein the terminology data and commentary data corresponding to the term data are provided. It is configured to have an explanation data storage means for storing.

According to this structure, the terminology generation device is
The generated term data and the commentary data corresponding to the terminology data are accumulated in the commentary data accumulating means, thereby constructing a database of the glossary extracted from the input natural language text data in the commentary data accumulating means.

The terminology generation device according to a fifth aspect is the terminology generation device according to any one of the first to fourth aspects, wherein the input text data is news manuscript data. It has a certain structure.

According to this structure, the glossary generating device is
By inputting the data of the news manuscript from the outside, in addition to general terms, difficult terms included in the news manuscript,
Generates explanation data for terms from new words and coined words.

A terminology generation program according to a sixth aspect of the present invention is a learning program in which input natural language text data and learning data that is a feature when the adnominal modifier is an explanatory sentence defining a term are registered. In order to generate the explanation data that defines the term in the text data from the database, the computer is configured to operate by the following means.

That is, dependency analysis means for generating dependency information between clauses of the text data by performing morphological analysis and syntactic analysis on the text data, and at least one of a noun and a noun phrase from the text data. Term data extracting means for extracting as term data, concept data extracting means for extracting concept data indicating a superordinate concept of the term data from the text data based on the dependency information and a specific paraphrase expression. Modification data extraction means for extracting, as modification data, a adnominal modification clause defining the term data, based on the received information and the learning data, and an explanation for defining the term data based on the concept data and the modification data. It was used as an explanation data generation means for generating data.

According to this configuration, the glossary generation program generates the dependency information between the clauses of the text data by performing the morphological analysis and the syntactic analysis on the text data by the dependency analysis means, and the term data is generated. From the text data, the character string that is a noun or a noun phrase is extracted as term data from the text data by the extraction means, and based on the dependency information and a specific paraphrase expression, the concept data extraction means extracts the text data from the text data. The concept data indicating the superordinate concept of the term data is extracted, and the adjunct modifier clause defining the term data is extracted as the modifier data by the modifier data extracting means based on the dependency information and the learning data, and the commentary data is extracted. By the generation means, based on the conceptual data and the modified data,
Explanation data that defines the term data is generated.

As a result, the glossary generation program shows the term data and the superordinate concept of the term data by performing the dependency analysis on the input natural language text data by morphological analysis or syntactic analysis. Concept data and modification data defining the term data are extracted from the text data, and commentary data defining the term data is generated based on the concept data and the modification data.

Further, a terminology search device according to a seventh aspect of the present invention includes a glossary generation device according to the fourth aspect of the present invention for generating commentary data for explaining the term data from text data, and an input means for inputting the term data. , Based on term data
The explanation data searching means for searching the explanation data stored in the explanation data storage means is provided.

According to this structure, the glossary search device is
The glossary generation device generates commentary data that explains the terminology data from the input text data, the terminology data is input by the input means, and the commentary data search means causes the commentary data storage means in the glossary generation device to be input. The accumulated commentary data is searched, and the search result is output by the output means. As a result, the glossary search device causes the glossary generation device that generates the commentary data that explains the terminology data from the input text data, to the glossary data generation device that responds to the query of the terminology data from the user. Has a user interface function of searching for and outputting the search result.

[0026]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described in detail below with reference to the drawings. First Embodiment: Configuration of Glossary Generation Device FIG. 1 is a block diagram showing the overall configuration of a glossary generation device according to the first embodiment of the present invention. As shown in FIG. 1, the terminology generation device 1 inputs a news manuscript, which is text data, and includes term data included in the news manuscript,
This is a device that generates commentary data that defines the term data.

The terminology generation device 1 includes a dependency analysis means 11, a term data extraction means 12, a concept data extraction means 13, a concept database 14, and a modified data extraction means 1.
5, a learning database 16, an explanation data generation means 17, and an explanation data storage means 18 are provided. The news manuscript is assumed to be input as text data from the external news manuscript database 3.
Further, the glossary generating device 1 specifies the range of news manuscripts from the news manuscript database 3 (for example, 200 manuscripts).
An input device (not shown) for performing only 0-year news manuscripts is connected to the outside.

The dependency analysis means 11 decomposes the news manuscript input based on the range specification of the news manuscript input from the input device (not shown) into bunsetsu units by morphological analysis and syntactic analysis, and the bunsetsu The character string and the dependency relation of the clause are generated as dependency information, and the term data extraction means 12, the concept data extraction means 13 and the modification data extraction means 15 are notified.

The dependency analysis means 11 identifies a morpheme, which is the smallest linguistic unit having a meaning, from a news manuscript by morpheme analysis, and analyzes bunsetsu such as noun phrases and verb phrases and their dependency relationships by syntactic analysis. Identify. The morphological analysis and the syntactic analysis can be realized by a known technique.

The term data extraction means 12 extracts a clause serving as a term from the clauses included in the dependency information, based on the dependency information generated by the dependency analysis means 11, and outputs it as term data. This term data is notified to the conceptual data extraction means 13, the modified data extraction means 15, and the explanation data generation means 17.

Here, when a phrase that is a term is extracted as a noun or a noun phrase, the phrase is recognized as a term. If there is no clause to be recognized as a term, the term data is output as a null character string.

In addition to this, it is also possible to extract a character string to be treated as a term from surface text information in the text data of a news manuscript input to the terminology generation device 1. For example, in text data of a news manuscript, by determining a noun phrase enclosed in a bracket or the like as an important part,
The term data extraction means 12 can extract a character string enclosed in a bracket as a term.

The concept data extracting means 13 uses the dependency information generated by the dependency analyzing means 11 and the term data extracted by the term data extracting means 12 to generate a phrase indicating a superordinate concept of the term data. Extract as conceptual data.

The conceptual data is registered in the conceptual database 14 together with the term data. At this time, if the same term data is already registered, it is possible to register a plurality of concept data, and if the concept data corresponding to the term data is also the same, the number of appearances of the concept data is changed. Update (increment).

Then, the conceptual data extracting means 13 reads the conceptual data corresponding to the term data from the conceptual database 14 and notifies the comment data generating means 17 of the conceptual data. At this time, when there are a plurality of concept data corresponding to the term data, the comment data generating means 17 is notified of the concept data having the largest number of appearances in the concept data.

The concept data is the term data paraphrased by a specific paraphrasing expression (specific expression) such as “to” or “called”. Here, the dependency information generated by the dependency analysis unit 11 is used as a material for determining that the paraphrase has been paraphrased.

That is, the arrow (→) of “bunsetsu A → bunsetsu B” indicates the dependency relation, and the bunsetsu A is the origin and the bunsetsu B.
Is the reference destination, the input character string is "(Terminology data) → to say → (Noun (phrase))" or "(Terminology data) → is called → (Noun (phrase)) When there is a relation, it is determined that this noun (phrase) indicates a superordinate concept of term data. FIG. 3 shows the result of extracting the conceptual data by this specific expression from the actual news manuscript.
For example, in the character string “expression of emoticon”, the term data is “emoticon” (D1) and the concept data is “expression” (D2).

The concept database 14 is a database in which term data and concept data are registered in association with each other.
The conceptual data extracting means 13 registers and refers to term data and conceptual data. This term data and concept data have a one-to-many relationship, and for one term data,
A plurality of conceptual data are registered. The number of appearances is given to each piece of concept data, and the concept data extracting means 13 updates and refers to the number of appearances. The concept database 14 is composed of storage means such as a hard disk.

The modification data extraction means 15 is based on the dependency information generated by the dependency analysis means 11, the term data extracted by the term data extraction means 12, and the learning data registered in the learning database 16. And extract the adnominal modifier that modifies the term data, determine whether the adnominal modifier defines the term data, and use the adjunct modifier as the modifier data based on the result of the determination. Notify 17

To determine whether or not the adnominal modifier is defining the term data, the modifier data extracting means 15 has two terms, a verb directly related to the term data of the adnominal modifier and a particle immediately before it. This is performed based on the learning database 16 in which the set is registered as learning data. Note that this learning data is a known thesaurus that is a synonym / synonym list in which similar verbs are registered at the same time and similar words are classified into a tree structure.
Is built on the learning database 16.

Therefore, the verb in the adnominal modifier related to the term data is “v”, and the verb set that belongs to the same group as this verb “v” is “vg” as the thesaurus.
1 ”, the verb set belonging to the parent node of the verb“ v ”is“ vg
2 ", the verb" v the particle just before the "to" p ", also each weighting factor for the similarity of the verb" v "of the verb set" vg1 "and" vg2 "" w _a "
When the number of times "w _b ", the verb set "vg" and the particle "p" appear in the learning data is "n (vg, p)" and its expected value is "e (vg, p)", solidarity An index value “weight (v, p)” for determining whether the modifier is a so-called term definition clause that modifies term data is defined as in Expression (1).

[0042]

[Equation 1]

In this equation (1), n (vg1, p)
When <e (vg1, p), the first term is 0, and
When n (vg2, p) <e (vg2, p), the second term is set to 0. When the index value weight (v, p) is larger than a preset threshold value, it is determined that the adnominal modifier is an explanatory sentence defining the term data.

As an experimental result, FIG. 4 shows a part of the result of extracting the term data from the news manuscript of June 2001. In this experiment, 15295 learning data are given, w _a = 0.67, w _b = 0.33, and the threshold value is 1.0, and the term definition clause is extracted.

The learning database 16 is a database in which a binary set of a verb directly related to the term data of the adnominal modifier and the particle immediately preceding it is registered as learning data as described in the modification data extracting means 15, and is a hard disk. And the like.

The explanation data generation means 17 is based on the term data extracted by the term data extraction means 12, the concept data extracted by the concept data extraction means 13, and the modified data extraction means 15, and the term data, The explanation data that defines the term data is generated.

The modified data extracted by the modified data extracting means 15 described above is not appropriate as a definition sentence because the sentence ends in the adnominal form of a verb as shown in FIG. Therefore, the commentary data generating means 17 uses the “qualification data (union modifier clause) + conceptual data (superordinate concept)” to
Generate commentary data. For example, in FIG. 3, the term data is “emoticon” (D1) and the concept data is “expression” (D1).
2) Then, in FIG. 4, the modification data corresponding to the "smiley" is "expresses emotions using symbols etc. for e-mail exchanged with a personal computer or a mobile phone" (D
When it is 3), the commentary data corresponding to the "emoticon" is the expression that expresses emotions by using symbols etc. for the e-mail exchanged by the personal computer or the mobile phone by connecting the modification data and the conceptual data. Becomes

Further, if the modification data of the term data exists, but the conceptual data does not exist, the commentary data generating means 17 determines whether the final morpheme obtained by morphological analysis of the term data itself is the conceptual data. To do. This determination can be made, for example, by preliminarily registering, in the concept database 14, as concept data, data that is likely to be a superordinate concept in the final morpheme.

Here, when the final morpheme is conceptual data, the commentary data is generated by "modification data (union modification clause) + final morpheme of term data". For example, in FIG. 4, the term data “Judiciary System Reform Promotion Law”
If there is no conceptual data for (D4), "Law", which is the final morpheme of the "Judiciary System Reform Promotion Law" (D4)
As a conceptual data, by adding it to the adnominal qualification section "Define the system for promoting the reform of the judicial system" (D5) of the term data, the explanation that "the law for defining the system for promoting the reform of the judicial system" is added. Generate data.

Further, the comment data generating means 17 has the modifier data of the term data, but when the concept data does not exist and the final morpheme thereof cannot be determined to be the concept data, the "modifier data (adnominal modifier clause) +" The commentary data is generated by "thing". For example, in FIG.
If there is no conceptual data for the term data "Red Data Book" (D6), "Red Data Book"
By adding "things" to the adnominal modification clause "(recording endangered wild birds)" (D6) in (D6),
It generates commentary data that is "a record of wild birds that are threatened with extinction." The generated commentary data is stored in the commentary data storage means 18 in combination with the term data.

The explanation data accumulating means 18 is an accumulation means for accumulating the term data and the explanation data output from the explanation data generating means 17 as a pair, and is composed of a hard disk or the like. The data accumulated here can be used as a database for a glossary search in which the commentary data can be referred to from the term data.

The configuration of the glossary generation device 1 according to the present invention has been described above based on the embodiment, but the present invention is not limited to this. For example, the commentary data storage means 18 is not provided, The explanation data generation means 17 may output the term data and the explanation data to the outside as data, and may have a form in which the storage means is provided outside.

Further, the concept data extracting means 13 does not include the concept database 14, and the concept data extracting means 13 is based on the dependency information input from the dependency describing means 11 and the term data input from the term data extracting means 12. The concept data may be extracted for each dependency information.

The input data of the glossary generator 1 is
The text is not limited to news manuscripts, and may be general text data. For example, by inputting a manuscript as text data from a database accumulating manuscripts of magazines, it is possible to output commentary data for the latest term data of magazines.

(First Embodiment: Operation of Glossary Generation Device) Next, the operation of the glossary generation device 1 will be described with reference to FIGS. 1 and 5 to 8. FIG. 5 is a flowchart showing the operation of the entire glossary generation device 1. FIG. 6 is a flowchart showing the operation of the conceptual data extracting means 13. FIG. 7 is a flowchart showing the operation of the modified data extracting means 15. FIG. 8 is a flowchart showing the operation of the explanation data generating means 17. First, the operation of the entire glossary generation device 1 will be described based on the flowchart of FIG.

First, a news manuscript input from the outside is decomposed into bunsetsu units by morphological analysis and syntactic analysis, and a character string of the bunsetsu and a dependency relation of the bunsetsu are generated as dependency information ( Step a1).

Then, based on the dependency information, it is determined from the clauses included in the dependency information that the clause is a noun or a noun phrase, and a clause serving as a term is extracted as term data (step a2). ).

Here, it is determined whether or not there is term data extracted in step a2 (step a3), and if there is no term data (No), the operation ends. On the other hand, if there is term data (Yes), the process proceeds to step a4.

Then, based on the dependency information and the term data, a phrase indicating a superordinate concept of the term data is extracted as concept data by a specific expression such as “to” or “called” (step). a4).

In addition, an adnominal modifier that modifies the term data based on the dependency information, the term data, and the learning data registered in the learning database,
It is determined whether or not the term data is defined, and the adnominal modifier is extracted as the modifier data based on the determination result (step a5).

Then, based on the concept data and the modification data, the commentary data defining the term data is generated (step a6).

Next, based on the flowchart of FIG.
The operation of the conceptual data extracting means 13 will be described. The flowchart is a detailed description of step a4 in FIG.

First, based on the inputted dependency information, in each clause of this dependency information, "to"
It is determined whether or not there is a specific expression such as "or called" (step b1), and if there is no specific expression (No)
Determines whether the input term data already exists in the concept database 14 (whether it is registered) (step b2), and if it does not exist in the concept database 14 (not registered) (No), The process ends assuming that there is no conceptual data corresponding to the term data. On the other hand, when the concept database 14 has (is registered) the concept data corresponding to the term data (Yes)
Goes to step b5.

If there is a specific expression in step b1 (Yes), concept data, which is a superordinate concept of term data, is extracted from the dependency relationship based on the specific expression (step b3).

Then, the term data and the concept data extracted in step b3 are registered in the concept database 14 (step b4). At this time, if the same term data is already registered, it is possible to register a plurality of concept data, and if the concept data corresponding to the term data is also the same, the number of appearances of the concept data is changed. Update (increment).

Then, among the concept data corresponding to the term data, the concept data having the highest appearance frequency based on the number of appearances is output as the concept data of the term data (step b5).

When the concept database 14 does not exist in the glossary generation device 1, when there is no specific expression (No) in the judgment of step b1, it is considered that the concept data corresponding to the term data does not exist. To finish. On the other hand, if there is a specific expression (Yes), in step b3, based on the specific expression, concept data that is a superordinate concept of the term data is extracted from the dependency relationship.
It is also possible to simplify the conceptual data extraction operation by outputting and ending the processing.

Next, based on the flowchart of FIG.
The operation of the modified data extracting means 15 will be described. The flowchart is a detailed description of step a5 in FIG.

First, the dependency information that has been input,
Based on the term data, it is determined whether or not there is a adnominal modifier related to the term data in each clause of the dependency information (step c1), and there is no adnominal modifier related to the term data (No). Ends the processing, assuming that there is no qualified data related to the term data.

On the other hand, in step c1, if there is a adnominal modifier related to the term data (Yes), the index value for judging whether or not the adnominal modifier is an explanatory sentence defining the term data is used in the above (1). It is calculated based on the formula (step c2).

Then, the index value is compared with a preset threshold value (step c3). If the index value is less than or equal to the threshold value (No), this adnominal modifier defines the term data. The process ends as there is no qualifying data. On the other hand, when the index value is larger than the threshold value (Ye
In s), it is determined that this adnominal modifier is the modifier data relating to the term data, and this adnominal modifier is output as the modifier data (step c4).

Next, based on the flowchart of FIG.
The operation of the explanation data generating means 17 will be described. The flowchart is a detailed description of step a6 in FIG.

First, it is judged whether or not there is modified data which is an explanatory sentence defining the term data (step d1). If there is no modification data (No), it is further determined whether there is concept data that is a superordinate concept of the term data (step d2). If there is no concept data (No), the comment data variable is set to ""None" (null data) is set (step d3), and the process proceeds to step d10. On the other hand, if there is conceptual data in step d2 (Yes
In s), the character string of "conceptual data" is set in the variable of the comment data (step d4), and the process proceeds to step d10.

In step d1, if there is modified data that is an explanatory sentence defining the term data (Yes
In step s), it is further determined whether there is concept data that is a superordinate concept of the term data (step d5). If there is conceptual data (Yes), the character string of "qualification data + conceptual data" is set in the variable of the commentary data (step d6), and the process proceeds to step d10.

On the other hand, in step d5, when there is no conceptual data which is a superordinate concept of the term data (No), it is judged whether or not the final morpheme of the term data is conceptual data (step d7), and the final morpheme is the concept. If it is data (Yes), the character string of “qualification data + final morpheme” is set in the variable of the comment data (step d8), and the process proceeds to step d10.

If the final morpheme of the term data is not the conceptual data in step d7 (No), the character string of "qualification data + thing (thing)" is set in the variable of the comment data (step d9). Go to step d10.

Then, the term data and the commentary data set by the above processing are output (step d10).
Through the above operation, the glossary generation device 1 can generate term data and commentary data defining the term data from a news manuscript.

The terminology generation device 1 can realize each function in a computer by a program, or can combine each function program and operate as a terminology generation program.

(Second Embodiment: Glossary Search Device) FIG. 2
FIG. 6 is a block diagram showing an overall configuration of a glossary search device according to a second embodiment of the present invention. As shown in FIG. 2, the glossary search device 2 inputs a news manuscript, which is text data, generates term data contained in the news manuscript, and commentary data defining the term data, and also an operator. Is a device that outputs commentary data for the term data for which an inquiry has been made.

The terminology search device 2 is configured by adding commentary data search means 21, input means 22 and output means 23 to the terminology generation device 1 shown in FIG. Explanation data retrieval means 21, input means 22 and output means 23
Structures other than the above are denoted by the same reference numerals as those shown in FIG.
The description is omitted. Further, the glossary search device 2 is externally connected with an input device 4 such as a keyboard and an output device 5 such as a CRT.

The explanation data search means 21 is an input means 22.
The comment data corresponding to the term data accumulated in the comment data accumulating unit 18 is searched based on the term data which is the text data input from the, and the searched comment data is output to the output unit 23 as the text data. To do.

The input means 22 inputs and analyzes the input data input by the operator from the input device 4 such as a keyboard and a mouse. When term data is input as this input data, the term data is used as the explanation data search means 2
Output to 1.

When the input data specifies a range of news manuscripts, for example, "2000 news manuscripts", the range specification data is notified to the terminology generation device 1 so that the terminology generation device is notified. 1 inputs the news manuscript within the range from the news manuscript database 3,
The term data is extracted, commentary data defining the terminology data is generated, and the terminology data and the commentary data are output as text data to the output means 23.

The output means 23 outputs the term data and the comment data generated by the glossary generation means 1 and the comment data output after the search by the comment data search means 21 to the CRT.
To the output device 5, etc. as output data.

FIG. 9 shows a display example 50 of the output device 5 connected to the glossary search device 2. The screen of FIG. 9 is an example of a screen of an application in which an operator inputs term data and displays term data of a news manuscript corresponding to the term data.

For example, when the operator inputs term data into the term input field 50a via the input device 4 and presses the RETURN key or clicks the search button 50b with the mouse, the glossary search device 2 The comment data corresponding to the term data accumulated in the comment data accumulating unit 18 is searched, and the comment data is displayed in the comment display field 50d. Here, the clear button 50c is a button for collectively deleting the commentary data entered in the term input field 50a, and the end button 50e is a button for ending the application for explaining the terminology of the news manuscript. .

As described above, the terminology search device 2 generates term data and commentary data defining the terminology data from the news manuscript, and outputs commentary data for the terminology data inquired by the operator. By having an interface, you can always quickly find the latest explanation of a term.

The terminology search device 2 according to the present invention has been described above based on the embodiment. However, the present invention is not limited to this. For example, the terminology search device 2 has a network interface and can be remotely accessed via a network. It is also possible for the local operator to retrieve the commentary data of the term data.

[0089]

As described above, the terminology generation device, the terminology generation program, and the terminology search device according to the present invention have the following excellent effects.

According to the first aspect of the present invention, the terminology generation device performs the dependency analysis on the input natural language text data by morphological analysis or syntactic analysis to obtain the term data, It is possible to extract concept data indicating a superordinate concept of the term data and modification data defining the term data from the text data, and generate commentary data defining the term data based on the concept data and the modification data. .

As a result, the term data that is the source of the terminology can be extracted from the input natural language text data, and the commentary data that is the definition sentence of the term data can be extracted at the same time. Therefore, even with a large amount of text data, the glossary data can be extracted at high speed without human intervention.

According to the second aspect of the present invention, the terminology generation device selects, from the plurality of concept data corresponding to the term data in the concept database, the concept data having the highest appearance frequency as the term data of the term data. Since it can be conceptual data, by adding this conceptual data to the end when generating the explanatory data, if the superordinate concept is the same term data, the explanatory data is composed of the same conceptual data at the end. Therefore, it is possible to generate uniform glossary data.

According to the third aspect of the present invention, the terminology generation device determines, from among a plurality of adnominal modifiers, an adjunct modifier that is a descriptive sentence defining a term, and a verb directly related to the term and its immediately preceding verb. Since it is specified by a combination with the particle of, it is possible to determine whether or not the adnominal modifier is a definition of a term, because it is not necessary to consider the meaning of the sentence. The determination can be performed at high speed.

According to the fourth aspect of the present invention, the terminology generation device stores the generated term data and the comment data corresponding to the term data in the comment data storage means. , It is possible to build a glossary of input natural language text data. This makes it possible to easily build a database of terminology, and it is also possible to build a terminology data search system using this database.

According to the fifth aspect of the invention, in the terminology generation device, since the input text data is the data of the news manuscript, especially in the news manuscript, difficult terms and many new words are created every year. Or, a word coined by combining existing words is often accompanied by an explanation of the term so that the viewer can easily understand it, and by using this explanation, the glossary data can be effectively generated.

According to the sixth aspect of the present invention, the terminology generation program performs the dependency analysis on the input natural language text data by morphological analysis or syntactic analysis. It is possible to extract conceptual data indicating a superordinate concept of the term data and modification data related to the term data from the text data, and generate commentary data defining the term data based on the concept data and the modification data.

As a result, the term data that is the source of the terminology can be extracted from the input natural language text data, and further, the commentary data that is the definition sentence of the term data can be extracted at the same time. Therefore, even with a large amount of text data, the glossary data can be extracted at high speed without human intervention.

According to the seventh aspect of the present invention, the glossary search device asks the glossary generation device for generating commentary data for explaining the term data from the input text data to the user for inquiring the term data. On the other hand, it is possible to provide a user interface for searching the commentary data corresponding to the term data and outputting the search result, and the user can always easily search the meaning of the latest term data.

[Brief description of drawings]

FIG. 1 is a block diagram showing an overall configuration of a glossary generation device according to an embodiment of the present invention.

FIG. 2 is a block diagram showing an overall configuration of a glossary search device according to an embodiment of the present invention.

FIG. 3 is a diagram showing a part of a conceptual data extraction result of term data using a specific expression of the present invention.

FIG. 4 is a diagram showing a part of a result of extraction of adnominal modification clauses defining term data of the present invention.

FIG. 5 is a flowchart showing an operation of the glossary generation device according to the exemplary embodiment of the present invention.

FIG. 6 is a flowchart showing an operation of extracting conceptual data according to the embodiment of the present invention.

FIG. 7 is a flowchart showing an operation of extracting decoration data according to the embodiment of the present invention.

FIG. 8 is a flowchart showing an operation of generating commentary data according to the embodiment of the present invention.

FIG. 9 is a diagram showing an example of an output screen of the glossary search device according to the embodiment of the present invention.

[Explanation of symbols]

1 ... Glossary generator 2 ... Glossary search device 11 ... Dependency analysis means 12 ... Term data extraction means 13 ... Conceptual data extraction means 14 ... Concept database 15 ... Modification data extraction means 16 ... Learning database 17 ... Explanation data generation means 18 ... Explanation data storage means 21 …… Explanation data search method 22 ... Input means 23 ... Output means

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Nobuyuki Yagi             1-10-11 Kinuta, Setagaya-ku, Tokyo, Japan             Broadcasting Association Broadcast Technology Institute F-term (reference) 5B009 MB03 ME25 MF02                 5B075 ND03 ND22 NK02 NR02 NR20                       UU01                 5B091 AA15 CA05 CC02 CC04 CC16

Claims (7)

[Claims] 1. A glossary generation device for generating commentary data defining term data from input natural language text data, the text data being subjected to morphological analysis and syntactic analysis. Dependency analysis means for generating dependency information of the phrase, and a term data extraction means for analyzing a character string that becomes a noun or a noun phrase from the text data and extracting as term data, the dependency information, and a term Conceptual data extracting means for extracting conceptual data indicating a superordinate concept of the term data from the text data based on a specific paraphrasing expression for paraphrasing data, and when the adnominal modifier clause becomes an explanatory text defining a term in advance Based on a learning database in which learning data that is a characteristic of the learning data is registered, and the dependency information and the learning data Modification data extraction means for determining whether the adnominal modifier related to the term data is the definition of the term data, and extracting the adnominal modifier that is determined to be the definition as modifier data, the conceptual data and the modifier data And a commentary data generating means for generating commentary data that defines the terminology data, and a glossary generating device. 2. A concept database for registering the term data and a plurality of concept data corresponding to the term data is provided, wherein the concept data extracting means selects from the plurality of concept data corresponding to the term data, the frequency of occurrence thereof. The terminology generation device according to claim 1, wherein one concept data corresponding to the term data is determined based on the terminology data. 3. The learning database registers, as learning data, a verb directly related to a term and an immediately preceding particle when the adnominal modifier is an explanatory text defining the term. The glossary generation device according to claim 1. 4. The glossary according to any one of claims 1 to 3, further comprising: commentary data storage means for storing the term data and commentary data corresponding to the term data. Generator. 5. The glossary generation device according to claim 1, wherein the text data is data of a news manuscript. 6. A term in the text data is defined from input natural language text data and a learning database in which learning data that is characteristic when an adnominal modifier is an explanatory sentence defining the term is registered. In order to generate the commentary data, the computer performs a morphological analysis and a syntactic analysis on the text data to generate dependency information between clauses of the text data. Dependency analysis means, from the text data, a noun Or term data extraction means for extracting a character string that becomes a noun phrase as term data, based on the dependency information and a specific paraphrase expression,
Conceptual data extraction means for extracting conceptual data indicating a superordinate concept of the term data from the text data, based on the dependency information and the learning data, the adnominal modifier related to the term data defines the term data. A modification data extraction unit that determines whether or not the adnominal modification clause determined to be the definition is defined as modification data, and generates commentary data that defines the term data based on the concept data and the modification data. Explanation A glossary generation program characterized by functioning as data generation means. 7. A term for generating commentary data, which is a descriptive sentence defining terminology data, from the input natural language text data, and further searching for commentary data corresponding to the terminology data from the input terminology data. A glossary search device, which generates commentary data for explaining term data from the text data, a glossary generation device according to claim 4, an input means for inputting the term data, and based on the term data, A glossary search device comprising: commentary data search means for searching commentary data stored in the commentary data storage means; and output means for outputting the search results.