JP2850952B2 - Document search method and apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document search method and apparatus, and more particularly to a document search method and apparatus for searching a document by expanding a related keyword from an input keyword using a thesaurus dictionary and a word co-occurrence database.

[0002]

2. Description of the Related Art Conventionally, this type of document search method and apparatus has been used for setting a search character string from an input document or keyword and searching for a document including the search character string.

An example of a conventional document search apparatus is disclosed in Japanese Patent Laid-Open Publication No. Hei.
No. 172966. The document search device described in this publication is a device that inputs a document and searches for a similar document from a document database, and includes a document input unit and a dependency analysis unit that analyzes a dependency structure of the input document. An index extraction unit for determining a sentence structure from the dependency analysis result, extracting an index from the sentence structure and assigning importance, and a document storage unit for storing the input document, the dependency analysis result, and the index extraction result. A thesaurus expansion unit that expands the index of the index extraction unit in a thesaurus dictionary, and a similar document search unit that determines the similarity between the input document and the stored document from the similarity of the index and the similarity of the dependency relationship. And a similar document output unit for outputting the searched similar document.

[0004] The thesaurus development unit provides a polysemy determination table, and determines the distinction between words having the same notation but different meanings according to the field of the document.

[0005]

A first problem of the above-mentioned prior art is that related keywords which are not described in the thesaurus dictionary cannot be used for searching. The reason is that only synonyms and synonyms are described in the thesaurus.

A second problem is that in thesaurus expansion using the polysemy determination table, words that have different meanings may be expanded as synonyms or synonyms, so that search accuracy is not improved. The reason is that the polysemy determination table cannot select semantically correct synonyms and synonyms if the field information of the document cannot be specified, and cannot determine if the field information is not described in the document. That's why.

An object of the present invention is to develop a word having a co-occurrence relationship with an input keyword (hereinafter referred to as a co-occurrence word) as a related word, in addition to a synonym or a synonym of the input keyword, and to further convert the related word into another input word. It is an object of the present invention to provide a document search method and apparatus in which a keyword is selected as a related keyword when it has a co-occurrence relationship, and a document is searched using the input keyword and the related keyword.

Another object of the present invention is to recursively input a selected related keyword as an input keyword,
It is still another object of the present invention to provide a document search method and apparatus in which a document can be searched by expanding related keywords.

[0009]

According to a first document search method of the present invention, a related word composed of synonymous / synonymous words and co-occurring words is developed from an input keyword using a thesaurus dictionary and a word co-occurrence database, and is expanded. The related word is referred to the word co-occurrence database, a related word having a co-occurrence relationship with another input keyword is selected as a related keyword, and a document is searched using the input keyword and the related keyword. And

A second document search method according to the present invention develops related words composed of synonymous / synonymous words and co-occurring words from input keywords using a thesaurus dictionary and a word co-occurrence database, Referring to the word co-occurrence database, a related word having a co-occurrence relationship with another input keyword is selected as a related keyword, and the selected related keyword is recursively input as an input keyword, and the input keyword and the related keyword are input. Is used to search for a document.

A first document retrieval apparatus according to the present invention comprises a keyword input unit for inputting a keyword, and a synonym / synonym and co-occurrence using a thesaurus dictionary and a word co-occurrence database based on the input keyword input by the keyword input unit. A related keyword selecting unit that expands a related word composed of words, refers to the word co-occurrence database from the expanded related words, and selects a related word that is co-occurring with another input keyword as a related keyword; A document search unit that searches for a document from a document database using the keyword and the related keyword.

A second document search device of the present invention provides a keyword input unit for inputting a keyword, and a synonym / synonym using a thesaurus dictionary and a word co-occurrence database based on the input keyword input by the keyword input unit. Expanding related words consisting of co-occurring words, referring to the word co-occurrence database from the expanded related words, selecting related words that are co-occurring with other input keywords as related keywords, and selecting the selected related keywords A recursive related keyword selection unit for recursively inputting as an input keyword, and a document search unit for searching for a document from a document database using the input keyword and the related keyword.

[0013]

Next, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a document search device according to the first embodiment of the present invention. The document search device according to the present embodiment includes a keyword input unit 1, a word co-occurrence database 2, a thesaurus dictionary 3, a related keyword selection unit 4, a document database 5, and a document search unit 6. Have been.

The keyword input unit 1 inputs one or more keywords as a search character string using a keyboard or the like.

The word co-occurrence database 2 is a database in which co-occurrence relation information indicating which words appearing in the document database 5 or other document databases co-occur with what other words and at what frequency are described. It is. FIG. 3 illustrates the contents of the word co-occurrence database 2, but the frequency is omitted in FIG.

The thesaurus dictionary 3 is a dictionary in which synonyms and synonyms are described (see FIG. 3).

The related keyword selection unit 4 develops one or a plurality of input keywords given from the keyword input unit 1 into co-occurrence words using the word co-occurrence database 2, and uses the thesaurus dictionary 3 to generate synonyms and synonyms. / Or expand to synonyms (hereinafter referred to as synonyms and synonyms). Hereinafter, co-occurrence words and synonyms / synonyms are collectively referred to as related words. Next, using the word co-occurrence database 2, the related keyword selecting unit 4 checks whether or not the related word developed from a certain input keyword co-occurs with another input keyword. Select a word as a related keyword.

The document database 5 is a database that stores digitized documents.

The document search unit 6 searches and outputs a document from the document database 5 using the input keyword input to the keyword input unit 1 and the related keyword selected by the related keyword selection unit 4.

Referring to FIG. 2, the processing of the related keyword selecting section 4 includes a related word developing step A1, a co-occurrence relation determining step A2, and a related keyword selecting step A3.
And a related word presence / absence determination step A4 and an input keyword presence / absence determination step A5.

Next, the operation of the thus configured document retrieval apparatus according to the first embodiment will be described with reference to FIGS. 1, 2 and 3 together with the document retrieval method according to the first embodiment. I will explain.

When an input keyword is input from the keyword input unit 1, it is supplied to the related keyword selection unit 4.

The related keyword selecting section 4 develops co-occurrence words from the respective input keywords using the word co-occurrence database 2 and develops synonyms and synonyms using the thesaurus dictionary 3 (step A1).

Next, the related keyword selecting unit 4 checks whether or not the developed related word co-occurs with any other input keyword using the word co-occurrence database 2 (step A2), and determines the co-occurrence relationship. A certain related word is selected as a related keyword (step A3).

Subsequently, the related keyword selecting section 4 judges whether or not other related words have been developed (step A4), and if so, returns control to step A2, and returns to steps A2 to A4. repeat.

When there are no more expanded related words, the related keyword selection unit 4 determines whether there is another input keyword (step A5), and if there is another input keyword, returns control to step A1. And steps A1 to
Repeat A5.

For example, referring to FIG. 3, when three words "car", "USA", and "sanctions" are given as input keywords, "automobile", "synonymous" as synonyms and synonyms of the input keyword "car" are given. "Wheel", "Caster", "Wagon"
Are extracted from the thesaurus dictionary 3. Also, three words “dumping”, “traffic accident”, and “insurance” are extracted from the word co-occurrence database 2 as co-occurrence words having a co-occurrence relationship with “car”. Of these seven related words, the related words "car" and "dumping" that co-occur with the other keywords "USA" or "sanctions"
Is selected from the word co-occurrence database 2 as a related keyword. Similarly, for the input keyword "USA"
“Canter Trade Representative” is selected as a related keyword, and “Canter Trade Representative” and “Dumping” are selected as related keywords for the input keyword “sanctions”.

The document retrieval unit 6 retrieves and outputs a document from the document database 5 using the input keyword input from the keyword input unit 1 and the related keyword selected by the related keyword selection unit 4.

As described above, in the document search method and apparatus according to the first embodiment, the co-occurrence word and the related word including the synonym / synonym are expanded from the input keyword using the word co-occurrence database 2 and thesaurus dictionary 3. Then, a related keyword having a co-occurrence relationship with another input keyword can be selected from the related words using the word co-occurrence database 2 and used for document search. Further, even for related words having polysemantic meanings, it is possible to use the word co-occurrence database 2 to select only related keywords that are co-occurring with other input keywords and limit the meaning to unique words. Therefore, related words that have the same notation but have different meanings can be excluded from the related keywords. For this reason, even when the field information is not described in the document, the retrieval accuracy of the document is significantly improved.

FIG. 4 is a block diagram showing a configuration of a document search device according to the second embodiment of the present invention. The document search device according to the present embodiment is different from the document search device according to the first embodiment shown in FIG. 1 in that the related keyword selection unit 4 is replaced by a recursive related keyword selection unit 4 ′. Only the point is different.

The recursive related keyword selection unit 4 'recursively processes the related keyword selected by the related keyword selection unit 4 in the same manner as the input keyword, and further expands the related keyword. However, in order to prevent the number of expanded related keywords from diverging, the recursive related keyword selection unit 4 'sets only the case where the related keywords are proper nouns to be subjected to recursive processing.

Referring to FIG. 5, the processing of the recursive related keyword selecting section 4 'is performed by the steps A3 and A4 of the related keyword selecting section 4 in the document search apparatus according to the first embodiment shown in FIG. The relevant keyword proper noun determination step B1 is inserted between the two.

FIG. 6 is a diagram exemplifying the contents of the word co-occurrence database 2 and thesaurus dictionary 3 in the document search device according to the second embodiment.

Next, the operation of the thus configured document search apparatus according to the second embodiment will be described with reference to FIGS. 4, 5 and 6 together with the document search method according to the second embodiment. The following description focuses on differences from the document search method and apparatus according to the first embodiment.

When the related keyword selected in step A3 is a proper noun (Yes in step B1), the recursive related keyword selection unit 4 'returns control to step A1 to perform the same processing as the input keyword recursively. To do.

For example, referring to FIG. 6, when three words "car", "USA" and "sanctions" are given as input keywords, "punishment" and "penalty" are synonyms and synonyms of "sanctions". The three words “” and “penalty” are extracted from the thesaurus dictionary 3. In addition, as co-occurrence words co-occurring with “sanctions”, four words of “Super Article 301”, “Canter Trade Representative”, “Economy”, and “Dumping” are extracted from the word co-occurrence database 2. Of these seven words, co-occurrence words having a co-occurrence relationship with one of the other input keywords "car" and "USA" are extracted from the word co-occurrence database 2, and "canter trade representative" and "dumping" are input. It is selected as a keyword related to the keyword “sanctions”. Next, three words of “Super Article 301”, “US”, and “sanctions” are extracted from the thesaurus dictionary 3 as synonyms and synonyms of the related keyword “Canter Trade Representative” which is a proper noun, and are input keywords.
The co-occurrence word “Super 301” which is co-occurring with any of “car”, “USA” and “sanctions” is extracted from the word co-occurrence database 2 as a further related keyword of the related keyword “Canter Trade Representative”.

The effect of the document search method and apparatus according to the second embodiment is that, when many related keywords cannot be obtained from the input keywords by using the word co-occurrence database 2 and thesaurus dictionary 3, recursively. By expanding the related keywords, sufficient related keywords can be obtained and the document can be searched.

As a modification of the document search method and apparatus according to the second embodiment of the present invention, the recursive related keyword selecting section 4 'is set to prevent the number of related keywords from diverging. Alternatively, only words that exceed the threshold of the co-occurrence frequency may be processed.

In such a modification of the document search method and apparatus according to the second embodiment, the related keyword selected in step A3 has another input keyword with a co-occurrence frequency exceeding the threshold set in step A2. If it co-occurs with the keyword, the process from step A1 is performed in the same manner as the input keyword.

The effect of such a modification of the document search method and apparatus according to the second embodiment is that many related keywords can be obtained from the input keywords using the word co-occurrence database 2 and thesaurus dictionary 3. If that is not possible, sufficient relevance keywords can be obtained and the document can be searched by recursively expanding co-occurrence words with high co-occurrence frequency as relevance keywords.

By the way, in the document search methods and apparatuses according to the first and second embodiments, co-occurrence words are developed from the input keywords using the thesaurus dictionary 3, but other related dictionaries such as a bilingual dictionary are used. To expand related words.

By creating the word co-occurrence database 2 from the document database 5, co-occurrence relation information adapted to the field of the document database 5 is used.
A co-occurrence word with higher relevance can be selected as a related keyword.

Further, by providing a document input unit and a keyword extraction unit having a morphological analysis function in place of the keyword input unit 1, a document can be searched using a document as an input.

Further, in the development from the input keyword to the related word, the expansion is performed as the co-occurrence word only when the value exceeds a threshold set by paying attention to the co-occurrence frequency of the word co-occurrence database 2. Only strong co-occurrence words can be selected as related keywords.

Further, when selecting a related keyword using the word co-occurrence database 2 from among the related words developed by the thesaurus dictionary 3 and the word co-occurrence database 2 from the input keyword, the selection condition is set to several other conditions. By narrowing down to those that co-occur with the input keyword,
A co-occurrence word with higher relevance can be selected as a related keyword.

[0047]

As described above, the first effect of the present invention is that the input keyword is expanded. As a result, the retrieval accuracy of the document is improved. The reason is that, in addition to the synonyms and synonyms of the input keyword, a co-occurrence word having a strong relation is added as a related keyword to the input keyword to search for a document.

The second effect is that, even for a related word having a ambiguous meaning, only the related keyword having a co-occurrence relationship with another input keyword is selected using the word co-occurrence database, and the meaning is unambiguous. Can be limited to
A related word that has the same notation but has a different meaning can be excluded from related keywords. As a result, the retrieval accuracy of the document is improved. The reason is that when a related word is expanded from an input keyword, only a co-occurring word having a co-occurrence relationship with another input keyword is selected as a related keyword, and is added to the input keyword and searched.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a document search device according to a first embodiment of the present invention.

FIG. 2 is a flowchart illustrating a process of a related keyword selecting unit in FIG. 1;

FIG. 3 is a diagram illustrating contents of a word co-occurrence database and a thesaurus dictionary in FIG. 1;

FIG. 4 is a block diagram showing a configuration of a document search device according to a second embodiment of the present invention.

FIG. 5 is a flowchart showing a process of a recursive related keyword selection unit in FIG. 4;

FIG. 6 is a diagram illustrating contents of a word co-occurrence database and a thesaurus dictionary in FIG. 4;

[Explanation of symbols]

 Reference Signs List 1 Keyword input unit 2 Word co-occurrence database 3 Thesaurus dictionary 4 Related keyword selection unit 4 'Recursive related keyword selection unit 5 Document database 6 Document search unit A1 Related word development step A2 Co-occurrence relation determination step A3 Related keyword selection step A4 Related Word presence determination step A5 Input keyword presence determination step B1 Related keyword proper noun determination step

Continuation of front page (56) References JP-A-2-56068 (JP, A) JP-A-3-252767 (JP, A) JP-A-7-56948 (JP, A) Iganji, Kimoto "Dynamic Associative Search Using Thesaurus-Introduction of Link Weight- ", Proc. Of the 44th Annual Meeting of the Information Processing Society of Japan (Early 1994), p. 105-106 (Hei 4-3-17) (58) Fields surveyed (Int. Cl. ⁶ , DB name) G06F 17/30 JICST file (JOIS)

Claims (6)

(57) [Claims] 1. A related word composed of synonymous / synonymous words and co-occurrence words is expanded from an input keyword using a thesaurus dictionary and a word co-occurrence database. A document search method, wherein a related word having a co-occurrence relationship with an input keyword is selected as a related keyword, and a document is searched using the input keyword and the related keyword. 2. A related word composed of synonymous / synonymous words and co-occurrence words is expanded from an input keyword using a thesaurus dictionary and a word co-occurrence database. A related word having a co-occurrence relationship with an input keyword is selected as a related keyword, the selected related keyword is recursively input as an input keyword, and a document search is performed using the input keyword and the related keyword. Document search method. 3. The document search method according to claim 2, wherein recursive input is performed as an input keyword only when the related keyword is a proper noun. 4. A keyword input unit for inputting a keyword, and a related word composed of a synonym / synonym and a co-occurrence word is developed from the input keyword input by the keyword input unit using a thesaurus dictionary and a word co-occurrence database, A related keyword selecting unit that selects, as a related keyword, a related word having a co-occurrence relationship with another input keyword by referring to the word co-occurrence database from the expanded related words, and a document using the input keyword and the related keyword A document search device comprising: a document search unit that searches for a document from a database. 5. A keyword input unit for inputting a keyword, and a related word composed of synonymous / synonymous words and co-occurring words is developed from an input keyword input by the keyword input unit using a thesaurus dictionary and a word co-occurrence database. A recursive method in which the related co-occurrence with another input keyword is selected as a related keyword by referring to the word co-occurrence database from the expanded related words, and the selected related keyword is recursively input as the input keyword. A document search device, comprising: a related keyword selection unit; and a document search unit that searches for a document from a document database using the input keyword and the related keyword. 6. The document search device according to claim 5, wherein the recursive related keyword selection unit recursively inputs as an input keyword only when the related keyword is a proper noun.