JP2011059748A - Keyword type determination apparatus, keyword type determination method and keyword type determination program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a keyword type determination apparatus capable of determining the type of a keyword, even with respect to an arbitrarily defined keyword type. <P>SOLUTION: The keyword determination apparatus includes first to third document acquisition parts 1 to 3 that acquire documents related to a keyword from among previously prepared first to third document sets 9 to 11; first to third keyword type score calculating parts 4 to 6, that calculate the keyword type scores representing the degree of each keyword type likeliness using information in the documents acquired by the document acquisition parts 1 to 3; an integrated keyword type score calculating part 7 that calculates an integrated keyword type score, based on the keyword type score obtained by one or multiple keyword type score calculating parts; and a keyword type determination part 8 that determines upper N keyword types (N is an integer of one or more) having higher calculated integrated keyword type scores as the keyword types for the keyword. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a technique for determining a keyword type for a keyword.

  Currently, there are many systems that retrieve information about a keyword when the keyword is entered. There are various types, and there are various types of specialized search systems such as news search, image search, and product search depending on the information handled. However, switching the specialized search system depending on the type of keyword is a cumbersome operation for the user, and a single search system performs an appropriate specialized search according to the type of keyword entered and outputs the result. It is desirable to do. For this purpose, it is necessary to determine the type of the keyword for the input keyword.

  As a method for determining a keyword type from a keyword, Non-Patent Document 1 assigns a type to an entity in a document in advance and stores it together with the document. A set of assigned entities is acquired and ranked, and a keyword type is determined based on the types and ranks of entities included in the set.

David Vallet and Hugo Zaragoza, "Inferring the Most Important Types of a Query: a Semantic Approach", Proceeding of 31st annual international ACM SIGIR conference on Research and development in information retrieval, 2008, pp. 857-858

  However, the conventional method has a problem that the keyword type cannot be arbitrarily determined. In Non-Patent Document 1, the entity type is directly used as the keyword type. However, since the entity type is assigned to the entity using a specific expression extraction method, it can be assigned by extracting a specific expression such as “person name”. The type can be used as a keyword type, but a type that cannot be extracted by a specific expression extraction method such as “celebrity” cannot be used as a keyword type.

  The present invention has been made in view of the above-described problems, and obtains a document related to a keyword, and determines a keyword type using information in the document, thereby arbitrarily determining a keyword type. It is another object of the present invention to provide a keyword type determination apparatus, method, and program capable of determining a keyword type.

  In order to solve the above problems, in the present invention, in a keyword type determination device that determines a keyword type for a keyword, a document acquisition unit that acquires a document related to the keyword from a document group prepared in advance, and a document Based on the keyword type score obtained by the keyword type score calculation means for calculating the keyword type score that represents the degree of uniqueness of each keyword type using the information in the information, and the single or plural keyword type score calculation means Comprehensive keyword type score calculating means for calculating a type score, and keyword type determining means for determining the top N keyword types (N is an integer of 1 or more) having the highest overall keyword type score as the keyword type for the keyword. .

  According to the present invention, since a document related to a keyword is acquired and the keyword type is determined using information in the document, the keyword type can be determined even for an arbitrarily defined keyword type.

It is a block diagram of the keyword type determination apparatus 100 in the Example of this invention. It is a flowchart which shows the operation | movement of the pre-processing of the keyword type determination apparatus 100 in the Example of this invention. It is a flowchart which shows the operation | movement of keyword type determination of the keyword type determination apparatus 100 in the Example of this invention. It is explanatory drawing which shows the example of the group of the keyword type and the related word with respect to it in the Example of this invention. It is explanatory drawing which shows the example of the concept base in the Example of this invention. It is explanatory drawing which shows the example of the keyword type field database in the Example of this invention. It is explanatory drawing which shows the example of the document acquired by the 1st document acquisition part in the Example of this invention. It is explanatory drawing which shows the example of the document field vector calculated by the 1st keyword type score calculation part in the Example of this invention. It is explanatory drawing which shows the example of the keyword type score calculated by the 1st keyword type score calculation part in the Example of this invention. It is explanatory drawing which shows the example which totaled the selection frequency of the document with respect to the click log in the Example of this invention. It is explanatory drawing which shows the example of the document acquired by the 2nd document acquisition part in the Example of this invention. It is explanatory drawing which shows the example of the keyword type score calculated by the 2nd keyword type score calculation part in the Example of this invention. It is explanatory drawing which shows the example of the document acquired by the 3rd document acquisition part in the Example of this invention. It is explanatory drawing which shows the example of the keyword type category database in the Example of this invention. It is explanatory drawing which shows the example of the thesaurus in the Example of this invention. It is explanatory drawing which shows the example of the keyword type score calculated by the 3rd keyword type score calculation part in the Example of this invention. It is explanatory drawing which shows the example of the comprehensive keyword type score calculated by the comprehensive keyword type score calculation part in the Example of this invention.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings, but the present invention is not limited to the following embodiments.

  FIG. 1 shows the configuration of a keyword type determination apparatus 100 in an embodiment corresponding to claim 8 of the present invention.

  In FIG. 1, the keyword type determination apparatus 100 includes a first document acquisition unit 1, a second document acquisition unit 2, a third document acquisition unit 3, a first keyword type score calculation unit 4, and a second keyword type score. A calculation unit 5, a third keyword type score calculation unit 6, an overall keyword type score calculation unit 7, a keyword type determination unit 8, a first document set 9 stored in the first document set database, and a second document The second document set 10 stored in the set database, the third document set 11 stored in the third document set database, the concept base 12, the keyword type field database 13, and the click log stored in the click log database 14, the keyword type category database 15, and the keyword type field database 13 And a keyword type field database constructing unit (not shown).

  The first to third document acquisition units 1 to 3, the first to third keyword type score calculation units 4 to 6, the total keyword type score calculation unit 7, the keyword type determination unit 8, and the keyword type field database (not shown). Each function to be described later of the construction means is achieved by, for example, a computer.

  The first document acquisition unit 1 constitutes a first document acquisition unit of the present invention, and acquires a document related to a keyword from the first document set 9.

  The second document acquisition unit 2 constitutes a second document acquisition unit of the present invention. The second document acquisition unit 2 refers to the click log 14 and is different from the keyword type determination device 100 from the second document set 10. A document selected as a search result by a user in a search system is acquired as a document related to a keyword.

  The third document acquisition unit 3 constitutes a third document acquisition unit of the present invention, and acquires a document related to a keyword from the third document set 11 in which a category is assigned to the document.

  The first keyword type score calculation unit 4 constitutes the first keyword type score calculation means of the present invention, and uses a part or all of the text of the document acquired by the first document acquisition unit 1, A document field vector representing a field of text is calculated with reference to the concept base 12, and each of the calculated document field vector and each of the keyword type field database 13 stored in advance with reference to the keyword type field database 13. The degree of association with the keyword type field vector is calculated, and the degree of association is used as the keyword type score.

  The second keyword type score calculation unit 5 constitutes a second keyword type score calculation unit of the present invention, and uses a part or all of the text of the document acquired by the second document acquisition unit 2, A document field vector representing a field of text is calculated with reference to the concept base 12, and each of the calculated document field vector and each of the keyword type field database 13 stored in advance with reference to the keyword type field database 13. The degree of association with the keyword type field vector is calculated, and the degree of association is used as the keyword type score.

  The third keyword type score calculation unit 6 constitutes a third keyword type score calculation unit of the present invention, and uses the document category acquired by the third document acquisition unit 3 to generate the keyword type category database 15. To calculate the degree of relevance according to the hierarchical distance between the document category and the keyword type category (a value representing the relationship between categories that can be obtained using the category hierarchy). Degree is the keyword type score.

  The total keyword type score calculation unit 7 constitutes a total keyword type score calculation unit of the present invention, and is based on the keyword type score obtained by the first to third keyword type score calculation units 4 to 6. Calculate the keyword type score.

  The keyword type determination unit 8 constitutes the keyword type determination unit of the present invention, and the keyword type for the keyword is selected from the top N keyword types having a high total keyword type score calculated by the total keyword type score calculation unit 7. Is determined.

  In this embodiment, three document acquisition units and three keyword type score calculation units are prepared. However, the present invention is not limited to this, and an arbitrary number can be prepared. When a plurality of document acquisition units are the same and the keyword type score calculation unit is different, the document acquisition units may be combined into one, or the plurality of keyword type score calculation units are the same and the document acquisition unit is different. In some cases, the keyword type score calculation unit may be combined.

  Next, the operation of the keyword type determination device 100 will be described. The present invention has two steps: a pre-processing step for creating a database and a step of actually determining a keyword type.

  FIG. 2 is a flowchart showing the preprocessing operation of the keyword type determination apparatus 100.

  First, a keyword type field database construction unit (not shown) generates a keyword type field vector representing a keyword type field with reference to the concept base 12 for related words of each keyword type as shown in FIG. The calculation is made (S101), and a set of keyword type and keyword type field vector is stored in the keyword type field database 13 (S102).

  FIG. 3 is a flowchart showing the keyword type determination operation of the keyword type determination apparatus 100.

  First, the first document acquisition unit 1 acquires a document related to a keyword from the first document set 9 (S201).

  The first keyword type score calculation unit 4 uses part or all of the text of the acquired document to calculate a document field vector representing the field of the text with reference to the concept base 12 (S202), and the keyword type Referring to the field database 13, the degree of association between the calculated document field vector and each keyword type field vector stored in the keyword type field database 13 is calculated, and the degree of association is used as a keyword type score (S203). ).

  The second document acquisition unit 2 refers to the click log 14 and selects a keyword selected from the second document set 10 as a search result by a user in a search system different from the keyword type determination device 100. (S204).

  The second keyword type score calculation unit 5 calculates a document field vector representing the field of the text by referring to the concept base 12 by using a part or all of the text of the acquired document (S205). Referring to the field database 13, the degree of association between the calculated document field vector and each keyword type field vector stored in the keyword type field database 13 is calculated, and the degree of association is used as the keyword type score (S206). ).

  The third document acquisition unit 3 acquires a document related to the keyword from the third document set 11 (S207).

  The third keyword type score calculation unit 6 refers to the keyword type category database 15 using the acquired document category, and the keywords stored in the acquired document category and keyword type category database 15. The degree of association is calculated according to the hierarchical distance from the type category, and the degree of association is used as the keyword type score (S208).

  The total keyword type score calculation unit 7 calculates a total keyword type score based on the keyword type scores obtained by the plurality of keyword type score calculation units 4 to 6 (S209).

  The keyword type determination unit 8 determines the top N keyword types with the highest total keyword type scores calculated by the total keyword type score calculation unit 7 as the keyword types for the keywords (S210).

  Next, the operation of the keyword type determination apparatus 100 will be described more specifically. In the following description, the keyword type of the keyword “ABC” is determined.

  The pre-processing operation will be specifically described with reference to the flowchart shown in FIG.

  In step S101, keyword type field database construction means (not shown) calculates a keyword type field vector representing the keyword type field for the related words of each keyword type with reference to the concept base 12. FIG. 4 shows examples of keyword types and related words for the keyword types. As shown in FIG. 5, for example, the concept base 12 is a database in which a concept vector representing a word concept is assigned to a word.

  In step S <b> 102, keyword type field database construction means (not shown) stores a set of keyword type and keyword type field vector in the keyword type field database 13. Here, it is conceivable that an average of concept vectors of related words for each keyword type is used as a keyword type field vector for the keyword type. The method of creating the keyword type field database is not limited to this. An example of the created keyword type field database 13 is shown in FIG.

  Next, the keyword type determination operation will be described in detail with reference to the flowchart shown in FIG.

  In step S <b> 201, the first document acquisition unit 1 acquires a document related to the keyword “ABC” from the first document set 9. Here, it is assumed that the top three documents including many keywords in the document title or body are acquired as related documents using a conventional document search method. The document acquisition method is not limited to this, and an arbitrary number of documents may be acquired, or a document whose title matches the keyword or a document including the keyword may be acquired as a related document. . An example of the acquired document is shown in FIG.

  In step S202, the first keyword type score calculation unit 4 calculates a document field vector with reference to the concept base 12 by using a part or all of the text of the document acquired in step S201. The text of each document is text, and the text is divided into words using a morphological analysis technique. For example, if there is a concept vector for each word in FIG. 5, the concept vector is acquired. An average of concept vectors obtained from each word is a text vector, and an average of text vectors obtained from each text is a document field vector. An example of the document field vector is shown in FIG. The calculation method of the document field vector is not limited to this, the document title is also used as text, the weight is changed according to the importance of the word, and the vectors included in the main cluster by clustering multiple vectors You may use only.

  In step S203, the first keyword type score calculation unit 4 refers to the keyword type field database 13 and the document field vector calculated in step S202 and each keyword type field vector stored in the keyword type field database 13. Is calculated as a keyword type score. For each keyword type, the cosine distance between the document field vector and the keyword type field vector is calculated as the degree of association, and is used as the keyword type score. An example of the keyword type score is shown in FIG. The keyword type score calculation method is not limited to this, and may be calculated based on the Euclidean distance or the like.

  In step S <b> 204, the second document acquisition unit 2 refers to the click log 14 and acquires a document related to the keyword from the second document set 10. The click log 14 is a database that records which documents are selected from the keywords and search results used by the user of the search system. FIG. 10 shows an example in which, for the click log 14, the same keyword and selected document set are tabulated as the number of document selections. Here, among the documents related to the keyword “ABC” of the click log, the documents with document numbers 1, 2, and 3 are acquired as the top three documents having the highest number of selections. The document acquisition method is not limited to this, and all documents selected one or more times may be acquired. An example of a document acquired by the second document acquisition unit 2 is shown in FIG.

  In step S205, the second keyword type score calculation unit 5 uses a part of the text of the document acquired by the second document acquisition unit 2, for example, the first sentence or the whole, and refers to the concept base 12 for the document. Calculate the field vector. The document field vector calculation method is the same as that in step S202. Note that the document field vector calculation method may be different from that in step S202.

  In step S206, the second keyword type score calculation unit 5 refers to the keyword type field database 13 and the document field vector calculated in step S205 and each keyword type field vector stored in the keyword type field database 13. Is calculated as a keyword type score. The method for calculating the keyword type score is the same as in step S203. The keyword type score may be calculated by a method different from step S203. An example of the keyword type score calculated by the second keyword type score calculation unit 5 is shown in FIG.

  In step S207, the third document acquisition unit 3 acquires, as a related document, a document having the keyword “ABC” as a document title from the third document set 11 in which a category is assigned to the document. The document acquisition method is not limited to this, and a plurality of documents having the keyword “ABC” in the document body may be acquired. An example of a document acquired by the third document acquisition unit 3 is shown in FIG.

  In step S208, the third keyword type score calculation unit 6 refers to the keyword type category database 15 and the categories assigned to the document acquired in step S207 and the keywords stored in the keyword type category database 15. The degree of association is calculated according to the hierarchical distance from the type category (a value representing the relationship between categories that can be obtained using the hierarchical structure of the category) to obtain a keyword type score. The keyword type category database 15 is a database in which the keyword type category is assigned to each keyword type as shown in FIG. 14, for example. Here, the minimum value of the number of times of moving from the document category to the upper or lower category before reaching the keyword type category of each keyword type using the thesaurus shown in FIG. The distance is set, and the reciprocal of the value obtained by adding 1 to the distance is set as the keyword type score. In the example of FIG. 15, the thesaurus has a tree structure in which only one upper category of all categories exists. However, the present invention is not limited to this, and a structure in which a plurality of upper categories exist may be used. Further, the method for calculating the distance is not limited to the above, and the distance may be determined according to the depth of the tree that is a common ancestor category. The method for calculating the keyword type score is not limited to the above, and a value obtained by subtracting a value proportional to the distance from 1 may be used as the keyword type score. The document category of the keyword “ABC” is “actor” and “talent”, and the distance between the keyword type “entertainer” and the keyword type category “entertainer” is 1, so the keyword type score of the keyword type “entertainer” is 1 / (1 + 1) = 0.5. The same calculation is performed for the remaining keyword types. An example of the keyword type score calculated by the third keyword type score calculation unit 6 is shown in FIG.

  In step S209, the total keyword type score calculation unit 7 calculates a total keyword type score based on the three keyword type scores calculated by the first to third keyword type score calculation units 4 to 6. Here, the weights for the keyword type scores calculated by the first, second, and third keyword type score calculation units 4, 5, and 6 are set to 0.3, 0.6, and 0.1, respectively. A value obtained by adding a value obtained by multiplying the score by the weight is defined as an overall keyword type score. The method of setting the weight is not limited to this, and an arbitrary value may be set. In addition to the input keyword, a set of a plurality of keywords and the corresponding keyword type is prepared as correct data, and the correct data The weight may be adjusted so that the keyword type can be correctly determined for as many keywords as possible. The method for calculating the total keyword type score is not limited to the above, and a value obtained by adding a value obtained by multiplying the keyword type score by the reciprocal of the ranking may be used. 9, 12, and 16, the overall keyword type score for the keyword type “celebrity” is 0.377 × 0.3 + 0.371 × 0.6 + 0.500 × 0.1 = 0.386. The rest is calculated in the same way. An example of the total keyword type score calculated by the total keyword type score calculation unit 7 is shown in FIG.

  In step 210, the keyword type determination unit 8 sets the top N keyword types having a high overall keyword type score as keyword types for the keywords. Here, N = 1. The method of setting the value of N is not limited to this, and may be a value other than 1, or may be dynamically changed according to the value of the total keyword type score. From FIG. 17, the keyword type with the highest total keyword type score value is “celebrity”, so the keyword type of the keyword “ABC” is determined to be “celebrity”.

  As described above, the process described in this embodiment acquires a document related to a keyword, and determines the keyword type using information in the document, so that the keyword type can be determined even for an arbitrarily defined keyword type. Can be determined.

  In addition, the comprehensive keyword type score calculation part 7 is not restricted to the said Example, Based on the keyword type score obtained by any one or two of the 1st-3rd keyword type score calculation parts 4-6. A total keyword type score may be calculated.

  The keyword type determination apparatus 100 can be realized by installing a program for executing the series of operations described above in a computer. It is also possible to store the program in a hard disk device connected to a computer used as a keyword type determination device, a portable storage medium such as a flexible disk or a CD-ROM, and install and execute the program on the computer. It is. It is also possible to store the program in a network server and download and install it from there.

  The present invention is not limited to the above-described embodiments and examples, and various modifications and applications can be made within the scope of the claims.

  The present invention can be used for developing a search system that performs an appropriate specialized search according to the type of an input keyword.

DESCRIPTION OF SYMBOLS 100 ... Keyword type determination apparatus 1 ... 1st document acquisition part 2 ... 2nd document acquisition part 3 ... 3rd document acquisition part 4 ... 1st keyword type score calculation part 5 ... 2nd keyword type score calculation part 6 ... 3rd keyword Type score calculation unit 7 ... Total keyword type score calculation unit 8 ... Keyword type determination unit 9 ... First document set 10 ... Second document set 11 ... Third document set 12 ... Concept base 13 ... Keyword type field database 14 ... Click log 15 ... Keyword type category database

Claims (10)

A keyword type determination device that determines a keyword type for a keyword,
Whether to search for documents from a set of documents prepared in advance, based on which document selected by the user in a search system, or to search for a set of documents with categories A document acquisition means for acquiring a document related to the keyword,
A keyword type score calculating means for calculating a keyword type score representing the degree of likelihood of each keyword type using information in a document, and a single or a plurality of texts that are information in the document acquired by the document acquiring means A keyword that represents a field of a keyword that is calculated in advance based on the calculated document field vector and a word concept vector that represents the concept of a word. The degree of association with the type field vector is calculated, and the calculated degree of association is used as a keyword type score, or the document category acquired by the document acquisition unit and the document to which the keyword type belongs for each keyword type. Calculate the relevance with the keyword type category categorized by field and calculate the relevance And keyword type score calculation means for the keyword type score,
An overall keyword type score calculating means for calculating an overall keyword type score based on a keyword type score obtained by a single or plural keyword type score calculating means;
Keyword type determination means for determining the top N keyword types (N is an integer of 1 or more) having the highest total keyword type score calculated by the total keyword type score calculation means as the keyword type for the keyword;
A keyword type determination device having A keyword type field vector representing the field is calculated for each keyword type with reference to a concept base in which a word concept vector representing a word and a concept corresponding to the word is stored, and the keyword type field vector together with the keyword type is calculated as a keyword type field. Has keyword type field database construction means to store in the database in advance,
The keyword type score calculating means calculates a document field vector representing a field of the document from a part or all of a single or a plurality of texts that are information in the document with reference to the concept base, and the keyword type field database The keyword type determination according to claim 1, wherein a degree of association between the calculated document field vector and each keyword type field vector in the keyword type field database is calculated, and the degree of association is used as a keyword type score. apparatus. The document acquisition means refers to a click log that records which keyword is used for a search by a user in a certain search system and which document is selected, and the user in a certain search system The keyword type determination apparatus according to claim 1, wherein the document selected as is acquired as a document related to the keyword. A keyword type category database in which a keyword type category, which is a category in which a document to which the keyword type belongs for each keyword type, is classified by field, is stored together with the keyword type,
The keyword type score calculation means refers to the keyword type category database and obtains a hierarchical distance between the document category acquired by the document acquisition means and the keyword type category (using a hierarchical structure of categories). The keyword type determination apparatus according to claim 1, wherein a degree of association is calculated according to a category), and the degree of association is used as a keyword type score. A keyword type field vector representing the field is calculated for each keyword type with reference to a concept base in which a word concept vector representing a word and a concept corresponding to the word is stored, and the keyword type field vector together with the keyword type is calculated as a keyword type field. Has keyword type field database construction means to store in the database in advance,
The document acquisition means includes a first document acquisition means for acquiring a first document related to a keyword from a document set prepared in advance, and a keyword and search result used by a user in a search system for a search. Second document acquisition means for acquiring, as a second document related to a keyword, a document selected as a search result by a user in a search system with reference to a click log that records which document has been selected Have
The keyword type score calculating means calculates a document field vector representing a field of a document from a part or all of a single or plural texts that are information in the first document with reference to the concept base, and A first keyword type that refers to the keyword type field database, calculates a relevance between the calculated document field vector and each keyword type field vector in the keyword type field database, and uses the relevance as a keyword type score A keyword calculating field, calculating a document field vector representing a field of the document from a part or all of a single or plural texts as information in the second document with reference to the concept base, and the keyword type field Referring to the database, the calculated document field vector and the keyword type field data Calculating a degree of relevance between each keyword type field vector in the base, the keyword type determination apparatus according to claim 1, wherein a second keyword type score calculating means for the keyword type score the relevance. A keyword type field vector representing the field is calculated for each keyword type with reference to a concept base in which a word concept vector representing a word and a concept corresponding to the word is stored, and the keyword type field vector together with the keyword type is calculated as a keyword type field. Keyword type field database construction means to be stored in advance in the database;
A keyword type category database in which a keyword type category, which is a category in which a document to which the keyword type belongs for each keyword type, is classified by field, is stored together with the keyword type;
The document acquisition means includes a first document acquisition means for acquiring a first document related to a keyword from a document set prepared in advance, and a document related to a keyword from a document set prepared in advance and assigned a category. A third document acquisition means for acquiring a third document to be
The keyword type score calculating means calculates a document field vector representing a field of a document from a part or all of a single or plural texts that are information in the first document with reference to the concept base, and A first keyword type that refers to the keyword type field database, calculates a relevance between the calculated document field vector and each keyword type field vector in the keyword type field database, and uses the relevance as a keyword type score With reference to the score calculation means and the keyword type category database, the hierarchical distance between the category as information in the third document and the keyword type category in the keyword type category database (using the hierarchical structure of categories) Table showing the relationship between categories Calculating a degree of association according to the value), the keyword type determination apparatus according to claim 1, further comprising a third keyword type score calculating means for the keyword type score the relevance. A keyword type field vector representing the field is calculated for each keyword type with reference to a concept base in which a word concept vector representing a word and a concept corresponding to the word is stored, and the keyword type field vector together with the keyword type is calculated as a keyword type field. Keyword type field database construction means to be stored in advance in the database;
A keyword type category database in which a keyword type category, which is a category in which a document to which the keyword type belongs for each keyword type, is classified by field, is stored together with the keyword type;
The document acquisition means refers to a click log that records which keyword is used for a search by a user in a certain search system and which document is selected, and the user in a certain search system Second document acquisition means for acquiring a document selected as a second document related to a keyword, and a third document acquiring a third document related to the keyword from a document set prepared in advance and assigned a category. 3 document acquisition means,
The keyword type score calculating means calculates a document field vector representing a field of a document from a part or all of a single or a plurality of texts as information in the second document with reference to the concept database, A second keyword type that refers to the keyword type field database, calculates a relevance between the calculated document field vector and each keyword type field vector in the keyword type field database, and uses the relevance as a keyword type score With reference to the score calculation means and the keyword type category database, the hierarchical distance between the category as information in the third document and the keyword type category in the keyword type category database (using the hierarchical structure of categories) Relationship between categories Calculating a degree of association according to the value) representing the keyword type determination apparatus according to claim 1, further comprising a third keyword type score calculating means for the keyword type score the relevance. A keyword type field vector representing the field is calculated for each keyword type with reference to a concept base in which a word concept vector representing a word and a concept corresponding to the word is stored, and the keyword type field vector together with the keyword type is calculated as a keyword type field. Keyword type field database construction means to be stored in advance in the database;
A keyword type category database in which a keyword type category, which is a category in which a document to which the keyword type belongs for each keyword type, is classified by field, is stored together with the keyword type;
The document acquisition means includes a first document acquisition means for acquiring a first document related to a keyword from a document set prepared in advance, and a keyword and a search result used by a user in a search system for a search. Second document acquisition means for acquiring, as a second document related to a keyword, a document selected as a search result by a user in a search system with reference to a click log that records which document has been selected A third document acquisition means for acquiring a third document related to the keyword from a set of documents prepared in advance and assigned a category,
The keyword type score calculating means calculates a document field vector representing a field of a document from a part or all of a single or plural texts that are information in the first document with reference to the concept base, and A first keyword type that refers to the keyword type field database, calculates a relevance between the calculated document field vector and each keyword type field vector in the keyword type field database, and uses the relevance as a keyword type score A keyword calculating field, calculating a document field vector representing a field of the document from a part or all of a single or plural texts as information in the second document with reference to the concept base, and the keyword type field Referring to the database, the calculated document field vector and the keyword type field data A second keyword type score calculating means for calculating a degree of association with each keyword type field vector in the base and using the degree of association as a keyword type score; and referring to the keyword type category database, the third document The degree of relevance according to the hierarchical distance between the category that is the information in the list and the keyword type category in the keyword type category database (a value that represents the relationship between categories that can be obtained using the hierarchical structure of categories) The keyword type determination apparatus according to claim 1, further comprising: a third keyword type score calculation unit that calculates the relevance level as a keyword type score. A keyword type determination method for determining a keyword type for a keyword,
Document retrieval means retrieves a document from a set of documents prepared in advance, or a document to which a category is assigned, based on which document is selected from the retrieval results by a user in a certain retrieval system A document acquisition step of acquiring a document related to the keyword depending on whether the set is searched;
The keyword type score calculating means is a keyword type score calculating step for calculating a keyword type score representing the degree of likelihood of each keyword type using information in the document, and the information in the document acquired by the document acquiring means A document field vector representing a document field is calculated using a part or all of a single or a plurality of texts, and is calculated in advance based on the calculated document field vector and a word concept vector representing a word concept. The degree of relevance with the keyword type field vector representing the field of the selected keyword is calculated, and the calculated degree of relevance is used as the keyword type score, or the document category acquired by the document acquisition means and each keyword type Keyword type category that categorizes the document to which the keyword type belongs by field And keyword type score calculation step of calculating the degree of association with, and keyword type score the relevance of the calculated,
An overall keyword type score calculating means for calculating an overall keyword type score based on the keyword type score obtained by the single or plural keyword type score calculating means;
A keyword type determining step in which the keyword type determining means determines the top N keyword types (N is an integer of 1 or more) having the highest total keyword type score calculated by the total keyword type score calculating means as the keyword type for the keyword; ,
A keyword type determination method having A keyword type determination program for causing a computer to function as each means according to claim 1.

Images