JPH077416B2 - Terminology automatic creation method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Outline] The present invention relates to an automatic terminology creation method for automatically creating a terminology from a document, applying term extraction rules to terms extracted from a document, extracting term candidates, and extracting the term candidates from these term candidates. For the purpose of creating a glossary or checking the sufficiency of an existing glossary by creating excess or deficiency information, a natural language analysis unit that extracts words from a document and this natural language analysis unit For terms, apply term extraction rules to collect term candidates, and collect term candidates collected by this term collection section to create a glossary or exclude existing terms from term candidates. After that, a consolidation / integration processing unit for performing consolidation / integration to create excess / deficiency information is provided, and a terminology integrated or excess / deficiency information organized by this consolidation / integration processing unit is output. To achieve.

[Industrial application field]

The present invention relates to an automatic terminology creation system for automatically creating a glossary from a document.

[Problems to be Solved by Conventional Techniques and Inventions]

The terminology plays a large role in technical documents, especially computer manuals. In the past, this glossary was created manually, so it was inefficient and often leaked.

The present invention applies a term extraction rule to terms extracted from a document to extract term candidates, and creates a glossary from these term candidates, or creates excess or deficiency information to check the sufficiency of an existing glossary. The purpose is to do.

[Means for solving the problem]

 Means for solving the problems will be described with reference to FIG.

In FIG. 1, a document library 1 is a library for storing documents, especially technical documents.

The natural language analysis unit 3 refers to the Japanese basic dictionary 3-1 and the Japanese analysis grammar 3-3 for the input document (especially technical document) to perform morphological analysis, syntactic analysis, and the like to determine a phrase. It is something to take out.

The term collecting unit 4 applies the term extraction rule 4-1 to the terms extracted from the document to collect term candidates.

The consolidation / integration processing unit 10 consolidates the term candidates to create a glossary, or excludes existing terms from the term candidates.
This is to consolidate and create excess / deficiency information.

[Action]

According to the present invention, as shown in FIG. 1, the natural language analysis unit 3 analyzes the input document to extract a term, and the term collection unit 4 applies the term extraction rule 4-1 to this term. The term integration is extracted and the consolidation processing unit 10 organizes and integrates the term candidates to create a glossary, or the existing terms are excluded from the term candidates to create excess / deficiency information. There is.

Therefore, it is possible to automatically create a glossary from a document. Also, if there is already a glossary (for example, a technical glossary), the user can specify this to create excess / deficiency information, so it is possible to check the sufficiency of the existing glossary. Become.

〔Example〕

Next, the configuration and operation of one embodiment of the present invention will be sequentially described in detail with reference to FIGS. 1 to 3.

In FIG. 1, a document library 1 stores and manages documents (especially technical documents) created (Japanese manuscript writing) using a Japanese document creation device.

The integrated management tool 2 for technical terms, as described below,
It is a tool for creating a glossary from documents and checking for a lack of existing glossaries.

The natural language analysis unit 3 refers to the Japanese basic dictionary 3-1 and the Japanese analysis grammar 3-3 to perform word-level morphological analysis and syntactic analysis on the document read out from the document library 1 one by one, to analyze words and phrases. It is something to take out.

The term collection unit 4 applies term extraction rules 4-1 to the terms extracted from the document to extract term candidates. At this time, meanings are also extracted. The extracted term candidates and meanings are stored as the term “a1” and the meaning “b1” in the work file 6, and the English basic dictionary 3-2 is further searched to translate the term “a1” into English “d1”. Is also stored.

The term extraction rule 4-1 is a rule for extracting a term candidate by applying it to a term extracted as a target of term extraction by the index item reference unit 5 to be described later from the terms extracted from the document, and is placed in the glossary. It is a rule that determines whether or not the term is appropriate for. In the rule as shown in FIG. 3 (a), for example, the term extraction rule 1 is as follows: (a) + case particle [and] + assistant particle: is + ... + auxiliary verb (“is”, “is” "," Da ", etc.) is a rule for extracting candidate terms. Specifically, it is a rule for extracting the "transmitter" from the document described on the right side of the rule number 1 in the term extraction example in FIG.

The index item reference unit 5 refers to a tag (for example, underline: a method of mechanically adding to the index or a method of manually adding) that indicates in advance that the document is an index item, and the term extraction target is referred to. To find the phrase. This is based on the hypothesis that the set of term candidates is a subset of the set of index items.

The matching checking unit 7 checks whether or not the term candidates collected by the term collecting unit 4 and stored in the work file 6 are present in the technical term collection 8 created in advance.

The technical terminology 8 is a collection of technical terms in technical documents, and is a common technical term 8-0 common to a certain technology.
And product-specific terms (1) 8-1, 8-2 specific to each product
It is configured to exclude these technical terms from the term candidates.

The consolidation / integration processing unit 10 supplements the information (meaning, English translation, etc.) of incomplete terms with respect to the term candidates extracted by the term collection unit 4 to consolidate and create a glossary, or the matching check unit 7 With respect to the term candidates that have been inspected as existing in the technical terminology collection 8, these are excluded / consolidated to create excess / deficiency information.

The term information output unit 11 stores the glossary or excess / deficiency information consolidated by the consolidation and integration processing unit 10 in the term file 12, prints the excess / deficiency information list 13 of term definitions, and the like.

The writer's check 14 indicates that the writer checks the glossary or term definition excess / deficient information list 13.

Check 15 by the examination section indicates that the examination section further performs a final check on the glossary posted in the document such as the instruction manual of the product.

The term registration processing function 16 is a function for registering as a technical terminology 8 in a file after the check 15 of the examination section is completed.

Next, the operation of the configuration of FIG. 1 will be described in detail with reference to FIG. 3 according to the order described in the flowchart of FIG.

In FIG. 2, the target document is input at the beginning of the beginning.
This means that a target document (particularly a technical document) from which a term is to be extracted is read from the FIG. 1 document library 1. At this time, when checking the validity of the glossary, specify that fact. If no designation is made, the process of creating a glossary is performed.

In the figure (a), the natural language analysis of the document is performed. This is the first
A sentence (particularly a technical document) read from the diagram document library 1 is sequentially cut out, and the basic Japanese dictionary 3-
1 and Japanese analysis grammar 3-3, this means extracting words and phrases by morphological analysis (part of speech etc.), syntactic analysis, and semantic analysis.

In the figure (b), term candidates are collected. This is because the term found by the term collection unit 4 in FIG. 1A as an index item by the tag (for example, an underline) by the index item reference unit 5 in FIG. for,
It means that the term extraction rule 4-1 is applied to extract the term candidates and store them in the work file 6. For example, applying the term extraction rule 1 in FIG. 3 (a) to the document described on the right side of the term extraction rule 1 in FIG.
Means to extract.

(C) in the figure determines whether or not there is a term unique to the product.
This means that it is determined whether the product-specific term is excluded at the start and the product-specific terminology is registered in advance. If YES,
(D) The subsequent processing is performed. In the case of NO, the processing after (e) is performed.

In the figure (d), a matching check is performed. This means that the term candidates collected in (b) are inspected for existence in the terminology specific to the specified product.

In (e) in the figure, excess / deficiency information is organized and integrated. this is,
The term candidates collected in (b) are consolidated to create a glossary, or the term candidates that have been examined in (d) and excluded from the terminology in the technical glossary 8 are consolidated to consolidate term definitions. It means creating missing information. These consolidations include deleting duplicates from the word candidates, performing sort processing, and further organizing terms that lack necessary information (for example, meanings of terms shown in work file 6 or incomplete English translations). Is to be consolidated and added to the complete one).

In the figure, (f) outputs term information. This is a term file that contains terminology or excess / deficiency information organized and integrated in (e).
It means that the information is stored in 12, or the excess / deficiency information list 13 of term definitions is printed as a form.

With the above processing, it is possible to collect terms from a document and automatically create a glossary, and extract information about terms that are lacking in the existing glossary in response to the specified glossary inspection. It is possible to print the excess / deficiency information list 13 and to check the sufficiency of the existing glossary.

FIG. 3 (a) shows an example of term extraction rules, and FIG. 3 (b).
Indicates a verb / auxiliary verb example in the term extraction rule, and
Figure (C) shows an example of the extraction. Here, (a) to (g) represent term candidates to be extracted, * 1 to * 6 represent verbs / auxiliary verbs described in * 1 to * 6 in Fig. 3 (b), and [] can be omitted. Represents

Fig. 3 (a) Term extraction rule 1 (a) + case particle [and] + assistant particle: ha + ... + auxiliary verb
^{When * 1} is applied to the document on the right side of rule number "1" in Fig. 3 (c), "to" and "is" correspond to underlined, so it corresponds to the first (a). The phrase "transmitter"
Is extracted as a term candidate. Similarly, with respect to the other term extraction rules 2 to 7, the term candidates as shown by using "line" in FIG. 3C are extracted, respectively. These extracted term candidates can be consolidated and created as a glossary. Further, regarding these extracted term candidates, those existing in the technical terminology 8 can be excluded / consolidated to create excess / deficiency information, and the adequacy of the technical terminology 8 can be checked.

〔The invention's effect〕

As described above, according to the present invention, the term extraction rule is applied to the terms extracted from the document to extract the term candidates, and these term candidates are organized and integrated to create a glossary.
Or, since a structure is created in which excess / deficiency information is created by excluding / consolidating those existing in the existing glossary (such as a technical glossary) from these term candidates, the appropriate terms are collected from the document. In addition to being able to automatically create a glossary, it is possible to automatically create excess / deficiency information and check the sufficiency of an existing glossary (such as a technical glossary).

[Brief description of drawings]

FIG. 1 is a configuration diagram of an embodiment of the present invention, FIG. 2 is a flowchart for explaining the operation of the present invention, and FIG. 3 is a term extraction rule / extraction example. In the figure, 1 is a document library, 3 is a natural language analysis unit, and 3-
1 is a basic Japanese dictionary, 3-3 is a Japanese parsing grammar, 4 is a term collecting section, 4-1 is a term extracting rule, 6 is a work file, 7 is a matching checking section, 8 is a technical terminology, and 10 is an organization. An integrated processing unit, 11 is a term information output unit, 12 is a term file, and 13 is an excess / deficiency information list of term definitions.

Claims (1)

[Claims] 1. A natural language analyzing unit (3) for extracting a word from a document and a phrase (1) extracted by the natural language analyzing unit (3) in a glossary automatic creating method for automatically creating a glossary from a document. Term extraction rule (4-
Term collection unit (4) that collects term candidates by applying 1)
And a consolidation and integration processing unit that consolidates the term candidates collected by the term collection unit (4) to create a glossary, or excludes existing terms from the term candidates and then consolidates and creates excess and deficiency information. (10), and an automatic terminology creation method characterized by being configured to output terminology or excess / deficiency information that has been consolidated and integrated by the consolidation and integration processing section (10).