JP2008123271A - Knowledge base search system and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a knowledge base search system and method allowing a user to conduct a knowledge search, with little construction effort, in related knowledge fields that the user is unaware of. <P>SOLUTION: The knowledge base search system includes a knowledge storage device 13 serving as a knowledge base for storing knowledge; a knowledge base registration device 11 for storing a management table that, for each piece of knowledge stored in the knowledge base, defines fields to which the knowledge belongs, the weight of the knowledge, the relevance weight between pieces of knowledge in the same knowledge field, and the relevance weight between the knowledge fields; and a knowledge base search device 12 which, using both knowledge serving as a starting point for a search in the field to which the knowledge relevant to a search request belongs and the management table, searches for either or both of relevant knowledge in the same filed as the knowledge serving as the starting point and relevant knowledge in a field different from the knowledge serving as the starting point. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a knowledge base search system and a knowledge base search method for searching related knowledge in a knowledge base.

  In knowledge-based search, if a search method that uses keywords and attributes added to knowledge is used, important knowledge may not be searched depending on the search key used. Various methods for searching for related knowledge by adding associations and association strengths have been proposed (for example, Patent Documents 1 to 4).

  For example, Patent Literature 1 discloses a search method for selecting information related to a search key based on a table in which the strength of association between each element of information is set by a numerical value or the like.

  Further, Patent Literature 2 discloses a search method that enables searching for related knowledge by associating knowledge at a point that represents a user's viewpoint and a viewpoint that represents an intention.

  In Patent Document 3, “values” existing in knowledge are classified by “concepts”, and “values” for the “concepts” are managed by “a hierarchical structure of concepts defined in advance”. There are a plurality of hierarchies, and different hierarchies are used depending on the purpose of retrieval. However, a method is disclosed in which knowledge having concepts in close hierarchies is searched for as strongly related knowledge.

  Patent Document 4 discloses a method of searching related knowledge by a similar word search in consideration of a genre intended by a search key.

JP-A-4-127273 (information processing apparatus) Japanese Patent Laid-Open No. 9-114847 (information processing apparatus) JP-A-6-187157 (Problem Solving Support System) Japanese Patent Laid-Open No. 2001-184358 (Information Retrieval Device Based on Category Factor)

  However, in the technique described in Patent Document 1, the relationship between information is taken into consideration, but the relationship between all information that needs to be defined extends to a scale proportional to the square of the number of information. Is not large and realistic. Further, in the technique described in Patent Document 2, since it is necessary to add relations between all the knowledges, the addition is enormous, and the labor is similarly large and not realistic.

  In the technique described in Patent Document 3, only knowledge under the same conceptual hierarchy can be searched, and knowledge search cannot be performed between different knowledge fields. Further, with the technique described in Patent Document 4, it is not possible to search for knowledge fields that are not related to the genre of the search key.

  The present invention has been made in view of the above, and it is an object of the present invention to obtain a knowledge base search system and a knowledge base search method that enable a knowledge search in a related knowledge field that is not conscious by a user with a small construction effort. To do.

  In order to achieve the above-described object, the knowledge base retrieval system according to the present invention includes a knowledge base in which knowledge is accumulated, and each knowledge accumulated in the knowledge base, the field to which knowledge belongs, the weight of knowledge, and the same Knowledge base registration device for storing a management table for defining a relation weight between knowledges within a knowledge field, a relation weight between knowledge fields, knowledge serving as a base point searched in a field to which knowledge related to a search request belongs, and the management table And the related knowledge in the same field as the base knowledge and the related knowledge in other fields related to the base knowledge are searched for related knowledge of one or both of them. And a knowledge base search device.

  According to the present invention, the knowledge to be searched is classified in the knowledge field, and each knowledge is given a weight, and the relation between the knowledge is defined only within the same field, and the relation between the knowledge fields is given a separate weight. Therefore, it is possible to construct a knowledge base retrieval system that enables knowledge retrieval in related knowledge fields that the user is not aware of with less effort compared to the prior art that adds relationships between all knowledge. .

  According to the present invention, there is an effect that it is possible to construct a knowledge base search system that enables a knowledge search in a related knowledge field that the user is not conscious of with little effort.

  Exemplary embodiments of a knowledge base search system and a knowledge base search method according to the present invention will be explained below in detail with reference to the drawings.

  FIG. 1 is a block diagram showing a configuration of a knowledge base search system according to an embodiment of the present invention. The knowledge base search system shown in FIG. 1 includes an input / output device 10, a knowledge base registration device 11, a knowledge base search device 12, and a knowledge storage device 13 that is a knowledge base. Although not shown, these devices operate under the control of a control device including a CPU, a ROM that stores a control program that defines the operation of the CPU, and a RAM that provides a working area for the CPU. It is supposed to be.

  That is, the input / output device 10 performs reception processing such as a search request and response from the user and output processing such as a search result presentation to the user and an inquiry in the search process under the control of a control device (not shown).

  For example, as shown in FIG. 2, the knowledge base search device 12 includes a knowledge search device 20 that performs a search in a specific knowledge field by a conventional method, and a relationship in the same knowledge field using a search result of the knowledge search device 20. The same field knowledge search device 21 for searching for knowledge and the related field knowledge search device 22 for searching for knowledge in the related knowledge field using the search result of the knowledge search device 20 are provided, and are controlled by a control device (not shown). For example, in the procedure shown in FIG. 5, the knowledge weight, the relation weight between knowledge, and the relation weight between knowledge fields defined in the management table (see FIG. 3 and FIG. 4) received from the knowledge base registration apparatus 11 are used. The knowledge stored in the knowledge storage device 13 is searched.

  The knowledge base registration device 11 has, for each knowledge stored in the knowledge storage device 13, a knowledge weight, a field to which the knowledge belongs, and a relation weight management table between knowledges in which the relation weight between knowledge within the same knowledge field is defined ( 3) and an inter-field relation weight management table (see FIG. 4) in which relation weights between knowledge fields are defined, and the user input from the input / output device 10 under the control of a control device (not shown). A management table corresponding to the knowledge related to the search request is provided to the knowledge base search device 12.

  Here, with reference to FIG. 3 and FIG. 4, two types of management tables stored in the knowledge base registration apparatus 11 will be described. FIG. 3 is a diagram illustrating an example of an inter-knowledge weight management table that manages the knowledge field and the inter-knowledge weight. As shown in FIG. 3, in the inter-knowledge weight management table 30, for “knowledge” as a base point, the “affiliation field” of the knowledge, the “knowledge weight” within the knowledge affiliation field, and the knowledge “Relationship weight between knowledge”, which is the degree of association with other knowledge within the field of the affiliation, is defined. In FIG. 3, “inter-knowledge relation weight” indicates three relation degrees of relation 1, relation 2 and relation 3, but the number actually defined is not limited to this.

  FIG. 4 is a diagram showing an example of an inter-field relation weight management table for managing inter-field relation weights of knowledge. As shown in FIG. 4, the inter-field relation weight management table 40 defines “inter-area relation weights” that are degrees of association with the related knowledge fields with respect to “affiliation fields (knowledge fields)”. In FIG. 4, the “inter-field relation weight” indicates three relation levels of relation 1, relation 2, and relation 3, but the number actually defined is not limited to this.

  Next, the search process of the knowledge base search device will be described with reference to FIGS. FIG. 5 is a flowchart for explaining the search processing procedure of the knowledge base search device 12. In FIG. 5, the step that is the name of the processing procedure is abbreviated as ST.

  The search request given to the input / output device 10 by the user always has a search specification of “specific knowledge field”. In addition, any of related knowledge search specification in the same field and related knowledge search specification in the related field Since there is a case in which one of the designations is included and a case in which both designations are included, when a control device (not shown) receives a user search request from the input / output device 10, The knowledge search device 20, the same field knowledge search device 21 and the related field knowledge search device 22 in the knowledge base search device 12 are activated in the following procedure, and the corresponding search processing is executed.

  Since a search for a “specific knowledge field” is always specified in the user search request, a control device (not shown) first activates the knowledge search device 20 and sends a user search request to the knowledge search device 20. give. The knowledge search device 20 searches a specific knowledge field specified in the received user search request by a conventionally used method such as a search method using an attribute added to a keyword or knowledge, and a search result. Is stored as a search result knowledge group A (x) (ST11).

  That is, the knowledge search device 20 searches for knowledge as a base point such as “knowledge” A1 to A3, B1 to B3, C1, C2, D1, and D2 in the inter-knowledge related weight management table 30. When the search processing (ST11) in the knowledge search device 20 is completed, the control device (not shown) determines whether or not search of related knowledge in the same knowledge field is specified in the user search request received from the input / output device 10 ( ST12), it is checked whether or not retrieval of related knowledge in the related knowledge field is designated (ST15).

  As a result, when none of the designations is included (ST12: No) (ST15: No), the search result knowledge group A (x), which is the base point knowledge, is given to the input / output device 10 to be presented to the user. All search processes are terminated, but either related knowledge search in the same field is specified (ST12: Yes) or related knowledge search in the related field is specified (ST15: Yes). If either one or both are designated, each search is performed. Here, it will be described that related knowledge search in the same field is performed first, and then related knowledge search in the related field is performed, but the order may be reversed.

  That is, when a search for related knowledge in the same knowledge field is specified in the user search request received from the input / output device 10 (ST12: Yes), the control device (not shown) activates the same field knowledge search device 21. The search result knowledge group A (x), the inter-knowledge relation weight management table 30 and the “threshold value” are given to the same field knowledge search device 21.

  The “threshold value” may be determined in advance, or may be a method specified by the user. In the case of adopting a method designated by the user, a control device (not shown) inquires and obtains the user when the user's search request does not include a “threshold value”.

  The same-field knowledge search device 21 uses the inter-knowledge relation weight management table 30 for each of the search result knowledge groups A (x), and sequentially checks the knowledge that has relations in the same knowledge field, so that “knowledge weight” × “search result” The "inter-knowledge weight from knowledge group A (x)" is calculated, and the knowledge group whose calculation result exceeds the "threshold value" is added to the search result knowledge group A (x) (ST13).

  When the search processing (ST13) in the same field knowledge search device 21 is completed, the control device (not shown) checks whether or not to further search related knowledge in the same knowledge field (ST14). The number of search processes in the same field knowledge search device 21 may be included in the search request by the user, or may be determined by inquiring the user from a control device (not shown) each time. If the specified number of times is one, or if the result of inquiring the user is that the related knowledge in the same knowledge field is not further searched, the determination in ST14 is negative (No), and ST15 Proceed to

  On the other hand, if the specified number is two times or more, or if the result of inquiring the user indicates that the related knowledge in the same knowledge field is further searched, the determination in ST14 is affirmative (Yes). Returning to ST13, causing the same field knowledge search device 21 to execute the search process is repeated one or more predetermined times, and when the search process in the same field knowledge search device 21 ends, the determination in ST14 is negative (No). The process proceeds to ST15.

  When the search for related knowledge in the related knowledge field is not specified in the user search request (ST15: No), the control device (not shown) adds the search result knowledge group A (x) to the above same knowledge field in the present example. The search result obtained by adding the search result of the related knowledge is provided to the input / output device 10 to be presented to the user, and all search processes are terminated. If specified (ST15: Yes), this time, the same field The search result knowledge group A (x), the inter-knowledge relation weight management table 30, the inter-field relation weight management table 40, and the “threshold value” are given to the knowledge search device 21.

  The “threshold value” here may be determined in advance as described above, or may be a method specified by the user. In the case of adopting a method designated by the user, a control device (not shown) inquires and obtains the user when the user's search request does not include a “threshold value”.

  First, the same-field knowledge search device 21 uses the search result knowledge group A (x) and the inter-field relation weight management table 40 to specify the knowledge field to which the knowledge field-related weight is attached. Next, the inter-knowledge related weight management table 30 is applied to the knowledge in the associated knowledge field, and “knowledge field related weight” × “knowledge weight possessed by knowledge in the related knowledge field” is calculated, Those whose calculation results exceed the “threshold value” are stored as the search result knowledge group B (x) (ST16).

  Then, when the search process (ST16) in the same field knowledge search device 21 is completed, the control device (not shown) starts the related field knowledge search device 22, and the search result knowledge group B ( x), an inter-knowledge related weight management table 30 and a “threshold value” are given. The “threshold value” here may be determined in advance as described above, or may be a method specified by the user. In the case of adopting a method designated by the user, a control device (not shown) inquires and obtains the user when the user's search request does not include a “threshold value”.

  The related field knowledge search device 22 uses the inter-knowledge relation weight management table 30 for each of the search result knowledge groups B (x), and sequentially searches for knowledge that has relations in the same knowledge field, and obtains “knowledge weight” × “search result”. "Inter-knowledge relation weight from knowledge group B (x)" x "inter-knowledge field relation weight" is calculated, and the knowledge group whose calculation result exceeds the "threshold value" is added to the search result knowledge group B (x) (ST17). .

  When the search processing (ST17) in the related field knowledge search device 22 is completed, the control device (not shown) checks whether or not to further search related knowledge in the related knowledge field (ST18). The number of times of search processing in the related field knowledge search device 22 here may be included in the search request by the user in the same manner as described above, or may be determined by inquiring the user from a control device (not shown) each time. If the specified number of times is one, or if it is answered that the related knowledge in the same knowledge field is not further searched as a result of inquiring to the user, the determination in ST18 is negative (No), and the search The result knowledge group A (x), the search result of the related knowledge added thereto, and the search result knowledge group B (x) are given to the input / output device 10 to be presented to the user, and all search processes are completed.

  On the other hand, if the specified number of times is two or more, or if the result of inquiring the user is that the related knowledge in the same knowledge field is further searched, the determination in ST18 is affirmative (Yes). Returning to ST17, letting the related field knowledge search device 22 execute the search process is repeated one or more predetermined times, and when the search process in the related field knowledge search device 22 ends, the determination in ST18 is negative (No) The search result knowledge group A (x), the search result of the related knowledge added thereto, the search result knowledge group B (x), and the search result of the related knowledge added thereto are given to the input / output device 10 to the user. To complete the search process.

  Next, in order to facilitate understanding of the significance of the above search processing, specific calculation examples in ST11, ST13, ST16, and ST17 are shown. It is assumed that all “threshold values” are “5”.

  When the field of knowledge that the user requested to search is “affiliation field A”, in ST11, it is assumed that the existing method is used to search for the knowledge “knowledge A2” as a base point. That is, the search result knowledge group A (x) = knowledge A2.

In ST13, the following calculation is performed on knowledge A1 and knowledge A3 associated as related knowledge in the same field as knowledge A2 using inter-knowledge relation weight management table 30.
Knowledge A1: Knowledge weight 4 × Inter-knowledge weight from knowledge A2 1.5 = 6
Knowledge A3: Knowledge weight 4 x Knowledge related weight from knowledge A2 1.1 = 4.4
Since the calculation result 6 of the knowledge A1 exceeds the threshold value 5 from this calculation result, the knowledge A1 is added to the search result knowledge group A (x).

In ST16, referring to the inter-relationship related weight management table 40, the knowledge field B associated with the knowledge field A becomes a search target, so the knowledge B1 to knowledge B3 defined in the inter-knowledge related weight management table 30 are as follows. Perform the calculation.
Knowledge B1: Knowledge field related weight 1.7 × knowledge weight 5 = 3.5
Knowledge B2: Knowledge field related weight 1.7 × knowledge weight 9 = 6.3
Knowledge B3: Knowledge field related weight 1.7 × knowledge weight 6 = 4.2
From this calculation result, since the calculation result 6.3 of the knowledge B2 exceeds the threshold value 5, the knowledge B2 becomes a related field search result. That is, the search result knowledge group B (x) = knowledge B2.

Since the search target in ST17 is the search result knowledge group B (x) = related knowledge B1 and B3 from the knowledge B2, the following calculation is performed on the knowledge B1 and knowledge B3 defined in the inter-knowledge relation weight management table 30. Do.
Knowledge B1: Knowledge weight 4 x Knowledge related weight 1.5 from Knowledge B 2 x Knowledge domain related weight 1.7 = 5.25
Knowledge B3: Knowledge Weight 6 × Inter-Knowledge Relationship Weight 1.1 from Knowledge B2 × Inter-Knowledge Field Relationship Weight 1.7 = 4.62
Since the calculation result 5.25 of the knowledge B1 exceeds the threshold value 5 from the calculation result, the knowledge B1 is added to the search result knowledge group B (x) that is the related field search result.

  Through the above search process, the user who has requested the search in the knowledge field A can use the related knowledge “knowledge A1” in the same field that he / she was not aware of in addition to the original knowledge “knowledge A2”. Related knowledge “knowledge B1, knowledge B2” is presented as a search result.

  As described above, according to this embodiment, it is possible to search not only related knowledge within the same field, but also related fields only by including a search specification of related fields in a search request specifying a specific knowledge field. You can also search for knowledge at

  In this case, in this embodiment, the knowledge to be searched is classified in the knowledge field, weight is given to each knowledge, and the relation between the knowledge is defined only in the same field, and the relation between the knowledge fields is different. Since the definition is given with relation weights, it is possible to search for knowledge in related knowledge fields that the user is not aware of with less effort than the techniques described in Patent Documents 1 and 2, which add relations between all knowledge. A knowledge base search system can be constructed.

  As described above, the knowledge base search system and the knowledge base search method according to the present invention are useful for constructing a knowledge base search system that enables a knowledge search in a related knowledge field that the user is not conscious of with little effort. is there.

It is a block diagram which shows the structure of the knowledge base search system by one embodiment of this invention. It is a block diagram which shows the structural example of the knowledge base search apparatus shown in FIG. It is a figure which shows an example of the knowledge related weight management table which manages the affiliation field and the knowledge related weight of knowledge memorize | stored in the knowledge base registration apparatus shown in FIG. It is a figure which shows an example of the inter-field related weight management table which manages the inter-field weight of the knowledge memorize | stored in the knowledge base registration apparatus shown in FIG. It is a flowchart explaining the search processing procedure of the knowledge base search apparatus shown in FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Input / output device 11 Knowledge base registration device 12 Knowledge base search device 13 Knowledge storage device (knowledge base)
DESCRIPTION OF SYMBOLS 20 Knowledge search device 21 Same field knowledge search device 22 Related field knowledge search device 30 Knowledge related weight management table 40 Field related weight management table

Claims (5)

A knowledge base that accumulates knowledge,
Knowledge base registration device for storing a management table that defines a field to which knowledge belongs, a weight of knowledge, a relation weight between knowledges in the same knowledge field, and a relation weight between knowledge fields for each knowledge accumulated in the knowledge base When,
Using the knowledge as a base point searched in the field to which the knowledge related to the search request belongs and the management table, in the other field related to the related knowledge in the same field as the knowledge as the base point and the knowledge as the base point A knowledge base search system comprising: a knowledge base search device that searches for one of the related knowledge or the related knowledge of both. The knowledge base search device includes:
Retrieving related knowledge in the same field as the base point of knowledge by evaluating the weight of knowledge defined in the management table and the relation weight between knowledges in the same knowledge field as the knowledge The knowledge base retrieval system according to claim 1. The knowledge base search device includes:
By evaluating the weight of knowledge defined in the management table, the relation weight between knowledge within the same knowledge field as the knowledge, and the relation weight between knowledge fields, the others related to the knowledge as the base point 2. The knowledge base search system according to claim 1, wherein related knowledge in the field is searched. For each knowledge accumulated in the knowledge base, preparing a management table that defines the field to which the knowledge belongs, the weight of knowledge, the relationship weight between knowledge within the same knowledge field, the relationship weight between knowledge fields,
Searching for knowledge as a base point in the field to which the knowledge related to the search request belongs;
Searching for related knowledge in the same knowledge field as the base knowledge by evaluating a weight of knowledge defined in the management table and a relation weight between knowledges in the same knowledge field A knowledge base retrieval method characterized by that. For each knowledge accumulated in the knowledge base, preparing a management table that defines the field to which the knowledge belongs, the weight of knowledge, the relationship weight between knowledge within the same knowledge field, the relationship weight between knowledge fields,
Searching for knowledge as a base point in the field to which the knowledge related to the search request belongs;
The knowledge weight in the same knowledge field as the knowledge weight defined in the management table and the relation weight between knowledge fields are evaluated, and the knowledge in other fields related to the base point knowledge is searched. And a process of
A knowledge base search method characterized by including:

Images