JP2002334107A - Device, method for searching information, information search program and recording medium in which the same program is stored - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a device, a method for searching information, an information search program and a recording medium in which the program is recorded by which a search request is made clear by not only search conditions consisting of keywords as the search request but adding a search viewpoint according to a search purpose of a user and a search result exactly depending on the search request of the user is obtained. SOLUTION: When the search conditions consisting of the keywords to restrict a range of search and search viewpoints based on the search purpose of the user are accepted as the search request from a browser 153 of a user terminal, a document data base 119 is searched based on the search conditions, document IDs conformable to the search conditions are acquired, a document evaluation data base 123 is searched based on the document IDs and the search viewpoints, the respective scores of the document IDs in the search viewpoints, the document IDs are arranged in the order of the scores, the documents for the specified numbers of documents from the one with the highest evaluation are returned to a search result transmitting part 125 to the browser 153 as the search results.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention searches for and outputs information conforming to a search request in response to a search request input from a user terminal, and ranks search results using a user's search history. For example, an information search apparatus and method for performing a content search performed in a computer network represented by the Internet and ranking search results using a user's search history, an information search program, and an information search program The present invention relates to a recording medium on which a program is recorded.

[0002]

2. Description of the Related Art In a conventional search system, a search keyword is input to obtain search results ranked in order of relevance.

[0003] This ranking is created by a method based on the frequency of appearance of keywords in the content, a method based on the frequency at which each content is referenced from other content, and a frequency based on the frequency at which each content is referenced by the user. Based on the method.

[0004] By creating search results ranked by such a method, search results are presented in response to a user's search request from a large amount of search results, and the burden on the search is reduced.

Conventionally, as a document search apparatus for ranking documents as search results using a search history of past users, information that a user has referred to a document is obtained.
Some documents are ranked based on this reference frequency and output as search results.However, this is not necessarily a document that the user referred to the document as it is, and it was once ranked higher. Since the ranked documents are easily referred to, there is a problem that the information of the user is biased for a specific document. As a solution to such a problem, the user actually refers to the document. There is a way to use direct evaluation later.

[0006]

Among the above-described ranking creating methods in the conventional search system, the method using the frequency of the keyword has a problem that the search result contains many contents different from the purpose of the user. . This is because although the method is basically based on the idea that “content similar to the search request is suitable for the search request”, the keyword input in the keyword-type search system is 2, It is thought that this is because there are only about three words.

A conventional method based on the frequency of being referred to by other contents is a method in a case where a search target has a large and naturally established reference relationship like WWW. When applied to a group, there is a problem that no effect is obtained.

Further, the conventional method based on the frequency of reference from the user has a problem that the evaluation is biased depending on the ease of reference of the document. That is, it is conceivable that the content including many keywords and including the keyword with high frequency has a higher reference frequency, and the content once evaluated higher is more likely to be referred to, and the evaluation may be biased.

Further, as described above, in the conventional method in which the user uses the direct evaluation after actually referring to the document, the reference is unconsciously acquired by the user by acquiring the document at the time of retrieval. On the other hand, the direct evaluation has a problem that a sufficient number cannot be generally obtained because the input load of the user is large.

In this case, although the evaluation of the document can be performed accurately, there is a problem that the number of evaluations themselves is reduced, and the search results cannot be ranked using the evaluation of the document.

Furthermore, in a document search apparatus that ranks documents using the search history of past users that are the subject of the present study, a new document has no past evaluation, and therefore has no evaluation value. , There is a problem that the opportunity for a new document to be presented to the user is excessively small and is buried.

The present invention has been made in view of the above,
Its purpose is to clarify the search request by adding not only search conditions consisting of keywords as search requests, but also search viewpoints according to the user's search purpose, and to provide search results that appropriately meet the user's search requests. An object of the present invention is to provide an information retrieval apparatus and method that can be obtained, an information retrieval program, and a recording medium on which the program is recorded.

[0013]

In order to achieve the above object, according to the present invention, there is provided an information processing apparatus for searching and outputting information conforming to a search request in response to a search request input from a user terminal. A search device, a search request including a search condition including a keyword for limiting a search range as a search request, and a search viewpoint for evaluating a priority of information based on a search purpose of a user; A document database that stores a document ID that specifies each document corresponding to a keyword that is a word to be searched, and an evaluation value that is the relevance of each document with respect to each of the search viewpoints, for each search viewpoint and for each document ID. Search the document database based on the document evaluation database stored in the search request and the search condition received by the search request receiving means, and match the search condition A document search unit for obtaining a document ID, and searching the document evaluation database based on the obtained document ID and a search viewpoint received by the search request receiving unit, and an evaluation value of each of the obtained document IDs in the search viewpoint And a search result returning means for determining a document to be output as a search result based on the obtained evaluation value and returning the document to the user terminal.

According to the first aspect of the present invention, when a search condition including a keyword for limiting a search range and a search viewpoint based on a user's search purpose are received as a search request, a document database is generated based on the search condition. Search for
A document ID that satisfies the search condition is obtained, a document evaluation database is searched based on the document ID and the search viewpoint, each evaluation value of the document ID in the search viewpoint is obtained, and a search is performed based on the evaluation value. Since the document to be output as a result is determined and returned to the user terminal, the search request can be transmitted more clearly by adding the search viewpoint to the search request, and the existence of documents having different purposes or completely unnecessary documents can be removed. It is possible to output an accurate search result that meets the requirements, and the user can easily reach the target document, and can present the search result to the user based on the evaluation of the user who has performed a search from the same search viewpoint in the past. . In addition, it is possible to acquire data indicating a user's preference from each viewpoint by an information collection mechanism such as a document evaluation database, thereby providing a better service to an individual or the entire user. .

[0015] The present invention described in claim 2 provides the present invention in claim 1.
In the invention described in the above, the behavior information including browsing or evaluation of the user with respect to the search result transmitted from the user terminal as a result of browsing the search result by the user at the user terminal to which the search result is returned by the search result returning unit is described. A user action information receiving unit that receives the document ID that is the target of the action and the search viewpoint, a search action database that stores the received document ID, action information, and the search view; and a search action database that is stored in the search action database. Analyzing a user operation on the document based on the information, calculating an evaluation value of the document in the search viewpoint based on the analysis result, and using the calculated evaluation value for each of the search viewpoint and the document ID in the document evaluation database. The point is to have a search behavior analysis updating unit that updates the evaluation value.

According to the second aspect of the present invention, the action information including the browsing or evaluation of the user with respect to the search result transmitted from the user terminal as a result of browsing the search result at the user terminal together with the document ID and the search viewpoint. Receiving, storing the document ID, the behavior information, and the search viewpoint in the search behavior database; analyzing the user's operation on the document based on each information stored in the search behavior database; Of the document evaluation database and the evaluation value for each document ID are updated with the calculated evaluation value. Therefore, the evaluation value stored in the document evaluation database always reflects the behavior of the user. , The search results retrieved from this document evaluation database always reflect the user's actions such as browsing and evaluation , It is possible to present the best search results that match the search purpose of user to user.

Further, the present invention according to claim 3 is an information retrieval method for retrieving and outputting information matching the retrieval request in response to a retrieval request input from a user terminal, wherein the retrieval request is a retrieval request. A document I that accepts a search condition consisting of keywords that limit the range and a search viewpoint that evaluates the priority of information based on the user's search purpose, and specifies each document in accordance with the keyword that is a word included in each document
D is searched based on the received search condition, a document ID that satisfies the search condition is obtained, and an evaluation value, which is the relevance of each document with respect to each of the search viewpoints, is evaluated for each search viewpoint. And searching the document evaluation database stored for each document ID based on the obtained document ID and the received search viewpoint, obtains each evaluation value of the obtained document ID in the search viewpoint, and obtains the obtained evaluation. The gist is to determine a document to be output as a search result based on the value and return the document to the user terminal.

According to the third aspect of the present invention, when a search condition including a keyword for limiting a search range and a search viewpoint based on a search purpose of a user are received as a search request, a document database is determined based on the search condition. Search for
A document ID that satisfies the search condition is obtained, a document evaluation database is searched based on the document ID and the search viewpoint, each evaluation value of the document ID in the search viewpoint is obtained, and a search is performed based on the evaluation value. Since the document to be output as a result is determined and returned to the user terminal, the search request can be transmitted more clearly by adding the search viewpoint to the search request, and the existence of documents having different purposes or completely unnecessary documents can be removed. It is possible to output an accurate search result that meets the requirements, and the user can easily reach the target document, and can present the search result to the user based on the evaluation of the user who has performed a search from the same search viewpoint in the past. . In addition, it is possible to acquire data indicating a user's preference from each viewpoint by an information collection mechanism such as a document evaluation database, thereby providing a better service to an individual or the entire user. .

According to a fourth aspect of the present invention, in the third aspect of the present invention, the search result transmitted from the user terminal as a result of the user browsing the search result at the user terminal to which the search result is returned is provided. Action information including browsing or evaluation of the user is received together with the document ID and the search viewpoint as the target of the action, and the received document ID, action information, and search viewpoint are stored in the search action database, and the search action database Analyzing the user's operation on the document based on each stored information,
The gist is that an evaluation value of the document in the search viewpoint is calculated based on the analysis result, and the evaluation value for each of the search viewpoint and the document ID in the document evaluation database is updated with the calculated evaluation value.

According to the fourth aspect of the present invention, the action information including the browsing or evaluation of the user with respect to the search result transmitted from the user terminal as a result of browsing the search result at the user terminal together with the document ID and the search viewpoint. Receiving, storing the document ID, the behavior information, and the search viewpoint in the search behavior database; analyzing the user's operation on the document based on each information stored in the search behavior database; Of the document evaluation database and the evaluation value for each document ID are updated with the calculated evaluation value. Therefore, the evaluation value stored in the document evaluation database always reflects the behavior of the user. , The search results retrieved from this document evaluation database always reflect the user's actions such as browsing and evaluation , It is possible to present the best search results that match the search purpose of user to user.

According to a fifth aspect of the present invention, there is provided an information search program for searching for and outputting information conforming to a search request in response to a search request input from a user terminal. A document ID that accepts a search condition consisting of keywords that limit the range and a search viewpoint that evaluates the priority of information based on the search purpose of the user, and specifies each document in accordance with the keyword that is a word included in each document Is searched based on the received search condition, a document ID that satisfies the search condition is obtained, and an evaluation value that is the relevance of each document with respect to each of the search viewpoints is calculated for each search viewpoint. The obtained document I is stored in the document evaluation database stored for each document ID.
A search is performed based on D and the received search viewpoint, an evaluation value of each of the acquired document IDs in the search viewpoint is obtained, and a document to be output as a search result is determined based on the obtained evaluation value. The point is to return it to the terminal.

According to the fifth aspect of the present invention, when a search condition including a keyword for limiting a search range and a search viewpoint based on a user's search purpose are received as a search request, a document database is determined based on the search condition. Search for
A document ID that satisfies the search condition is obtained, a document evaluation database is searched based on the document ID and the search viewpoint, each evaluation value of the document ID in the search viewpoint is obtained, and a search is performed based on the evaluation value. Since the document to be output as a result is determined and returned to the user terminal, the search request can be transmitted more clearly by adding the search viewpoint to the search request, and the existence of documents having different purposes or completely unnecessary documents can be removed. It is possible to output an accurate search result that meets the requirements, and the user can easily reach the target document, and can present the search result to the user based on the evaluation of the user who has performed a search from the same search viewpoint in the past. . In addition, it is possible to acquire data indicating a user's preference from each viewpoint by an information collection mechanism such as a document evaluation database, thereby providing a better service to an individual or the entire user. .

Further, the present invention described in claim 6 provides the present invention according to claim 5.
In the invention described in the above, the behavior information including the browsing or evaluation of the user with respect to the search result transmitted from the user terminal as a result of the user browsing the search result at the user terminal to which the search result is returned is a target of the behavior. Received together with a certain document ID and the search viewpoint, store the received document ID, action information, and search viewpoint in a search action database, and analyze a user operation on the document based on each information stored in the search action database. The gist is to calculate an evaluation value of the document in the search viewpoint based on the analysis result, and to update the evaluation value for each of the search viewpoint and the document ID in the document evaluation database with the calculated evaluation value.

According to the present invention, action information including browsing or evaluation of the user with respect to the search result transmitted from the user terminal as a result of browsing the search result at the user terminal is recorded together with the document ID and the search viewpoint. Receiving, storing the document ID, the behavior information, and the search viewpoint in the search behavior database; analyzing the user's operation on the document based on each information stored in the search behavior database; Of the document evaluation database and the evaluation value for each document ID are updated with the calculated evaluation value. Therefore, the evaluation value stored in the document evaluation database always reflects the behavior of the user. , The search results retrieved from this document evaluation database always reflect the user's actions such as browsing and evaluation , It is possible to present the best search results that match the search purpose of user to user.

According to a seventh aspect of the present invention, there is provided a recording medium storing an information search program for searching for and outputting information conforming to a search request in response to a search request input from a user terminal. As a search condition consisting of keywords that limit the scope of the search, and a search viewpoint that evaluates the priority of information based on the search purpose of the user,
A document database that stores a document ID that specifies each document corresponding to a keyword that is a word included in each document is searched based on the received search condition, and a document ID that satisfies the search condition is obtained. A search is performed on a document evaluation database in which an evaluation value, which is the relevance of each document for each search viewpoint, is stored for each search viewpoint and for each document ID based on the acquired document ID and the received search viewpoint. Acquiring an evaluation value of each of the acquired document IDs from a viewpoint, determining a document to be output as a search result based on the acquired evaluation value, and recording an information search program to be returned to the user terminal on a recording medium. Make a summary.

According to the present invention, when a search condition including a keyword for limiting a search range and a search viewpoint based on a user's search purpose are received as a search request, a document database is determined based on the search condition. Search for
A document ID that satisfies the search condition is obtained, a document evaluation database is searched based on the document ID and the search viewpoint, each evaluation value of the document ID in the search viewpoint is obtained, and a search is performed based on the evaluation value. Determine the document to be output as a result, because the information search program to be returned to the user terminal is recorded on a recording medium, using the recording medium,
Its distribution can be improved.

The present invention described in claim 8 provides the present invention according to claim 7.
In the invention described in the above, the behavior information including the browsing or evaluation of the user with respect to the search result transmitted from the user terminal as a result of the user browsing the search result at the user terminal to which the search result is returned is targeted for the behavior Received together with a certain document ID and the search viewpoint, store the received document ID, action information, and search viewpoint in a search action database, and analyze a user operation on the document based on each information stored in the search action database. An information search program that calculates an evaluation value of the document in the search viewpoint based on the analysis result and updates the evaluation value for each of the search viewpoint and the document ID in the document evaluation database with the calculated evaluation value is a recording medium. It shall be recorded in the summary.

According to the present invention, action information including browsing or evaluation of the user with respect to the search result transmitted from the user terminal as a result of browsing the search result at the user terminal is recorded together with the document ID and the search viewpoint. Receiving, storing the document ID, the behavior information, and the search viewpoint in the search behavior database; analyzing the user's operation on the document based on each information stored in the search behavior database; The information retrieval program that updates the search viewpoint of the document evaluation database and the evaluation value for each document ID with this calculated evaluation value is recorded on a recording medium. Distribution can be improved.

Further, according to the present invention, a document obtained by searching a set of a plurality of documents from a document storage unit in response to a search request input from a user terminal is referred to in the user terminal, and From the search behavior at the user terminal performing the evaluation, an evaluation for the document is obtained, an evaluation value for each document is calculated based on the obtained evaluation, and a search result for a search request from the user terminal is calculated using the evaluation value. An information search apparatus for ordering outputs, wherein the search request, information for specifying the document, and a document evaluation acquisition unit for obtaining an evaluation for the document from a search action in a user terminal, the search request, document identification information, Document evaluation log storage means for storing evaluations for documents and documents, and direct evaluation for documents directly evaluated at the user terminal. Direct evaluation means for calculating an indirect evaluation of a document using a search request from the user terminal and information of the evaluation of the document obtained from the user terminal with respect to the document which is a search result corresponding to the search request. Evaluation means, document evaluation management means for calculating an overall evaluation value of each document from the direct evaluation and indirect evaluation of each document calculated by the direct evaluation means and indirect evaluation means, and the comprehensive evaluation value corresponding to each document And a document evaluation storage unit for storing the data.

According to the ninth aspect of the present invention, a search request, document identification information, and an evaluation of a document are obtained from a search action at a user terminal, and a direct evaluation of the document directly evaluated at the user terminal is calculated. Then, for a search request from the user terminal and for a document that is a search result for the search request, an indirect evaluation of the document is calculated using information of the evaluation of the document obtained from the user terminal. Calculates the comprehensive evaluation value of each document from the direct evaluation and indirect evaluation of the document, and stores the comprehensive evaluation value corresponding to each document. Indirect evaluation can be performed from the viewpoint of the evaluation, and a search request given from the user terminal to each document and a past user terminal The evaluation value given by using the evaluation value can be used to rank and output the documents required by the user terminal from the collection of documents, and when there are not enough evaluations to perform accurate evaluation on the document, When there is a document that does not have an evaluation such as a new document, it is possible to use the existing evaluation and the user's search log to identify the part of the directly evaluated document that the user would have actually evaluated. It is possible to increase the number of evaluations of a document without deteriorating the quality of the document evaluation.

According to a tenth aspect of the present invention, in the ninth aspect of the present invention, the indirect evaluation means obtains, from the user terminal, a search request from the user terminal and a document which is a search result corresponding to the search request. A partial document acquisition unit configured to acquire a partial document of interest in the user terminal from a directly evaluated document, which is a document directly evaluated in the user terminal, using the information on the evaluation of the document; A document relevance determining means for determining a document having high relevance, and an indirect evaluation of the document from the directly evaluated document based on the direct evaluation value of the directly evaluated document and the relevance of a certain document to the directly evaluated document. The point is to have an indirect evaluation management means to acquire.

According to the tenth aspect of the present invention, a partial document that is noticed by a user terminal is obtained from a directly evaluated document by using a search request and information of document evaluation, and the relevance of the partial document to the partial document is obtained. In order to determine the high document and obtain the indirect evaluation of the document from the directly evaluated document from the direct evaluation value of the directly evaluated document and the relevance of a certain document to the directly evaluated document, an accurate When there are not enough evaluations to perform the evaluation, or when there is a document that does not have an evaluation such as a new document, the existing evaluation and the user's search log are used to evaluate the actual Thus, it is possible to identify a portion that would have been evaluated by the user, and to increase the number of document evaluations without deteriorating the quality of the document evaluation.

According to the present invention, a document obtained by searching a set of a plurality of documents from a document storage unit in response to a search request input from a user terminal is referred to in the user terminal, and From the search behavior at the user terminal performing the evaluation, obtain the evaluation for the document,
An information search method that calculates an evaluation value for each document based on the obtained evaluation, and uses the evaluation value to order the output of search results in response to a search request from a user terminal. The information to be specified and the evaluation for the document are obtained from the search behavior in the user terminal, the search request, the document identification information, and the evaluation for the document are stored in the document evaluation log storage unit, and the evaluation is directly performed in the user terminal. Calculate the direct evaluation for the document, calculate the indirect evaluation of the document using the search request from the user terminal and the information of the evaluation of the document obtained from the user terminal for the document that is the search result for the search request, A total evaluation value of each document is calculated from the calculated direct evaluation and indirect evaluation of each document, and the total evaluation value is evaluated for each document. And summarized in that stored in the storage means.

According to the eleventh aspect of the present invention, a search request, document identification information and an evaluation of a document are obtained from a search action at a user terminal, and a direct evaluation of the document directly evaluated at the user terminal is calculated. Then, for a search request from the user terminal and for a document that is a search result for the search request, an indirect evaluation of the document is calculated using information of the evaluation of the document obtained from the user terminal. Calculates the comprehensive evaluation value of each document from the direct evaluation and indirect evaluation of the document, and stores the comprehensive evaluation value corresponding to each document. The indirect evaluation can be performed from the viewpoint of the evaluation, and a search request given from a user terminal to each document for a group of a plurality of documents and a past user terminal can be obtained. When the user terminal can rank and output the documents requested by the user terminal from the collection of documents using the evaluation value given by using these evaluations, and there are not enough evaluations to perform accurate evaluation on the documents If there is a document that does not have an evaluation such as a new document or a new document, use the existing evaluation and the user's search log to identify the part of the directly evaluated document that would have been evaluated by the user. And the number of document evaluations can be increased without lowering the quality of the document evaluation.

According to a twelfth aspect of the present invention, in the invention of the eleventh aspect, the processing for calculating the indirect evaluation is performed on a search request from a user terminal and a document which is a search result corresponding to the search request. Using the information on the evaluation of the document obtained from the user terminal, obtain a partial document that is noticed at the user terminal from the directly evaluated document, which is a document directly evaluated at the user terminal, and associate the partial document with the partial document. Determining a highly likely document, and obtaining an indirect evaluation of the document from the directly evaluated document from the direct evaluation value of the directly evaluated document and the relevance of a certain document to the directly evaluated document. I do.

According to the twelfth aspect of the present invention, a partial document that is noticed by a user terminal is acquired from a directly evaluated document by using a search request and information of document evaluation, and the relevance of the partial document to the partial document is obtained. In order to determine the high document and obtain the indirect evaluation of the document from the directly evaluated document from the direct evaluation value of the directly evaluated document and the relevance of a certain document to the directly evaluated document, an accurate When there are not enough evaluations to perform the evaluation, or when there is a document that does not have an evaluation such as a new document, the existing evaluation and the user's search log are used to evaluate the actual Thus, it is possible to identify a portion that would have been evaluated by the user, and to increase the number of document evaluations without deteriorating the quality of the document evaluation.

According to a thirteenth aspect of the present invention, the user terminal refers to a document obtained as a result of searching a set of a plurality of documents from the document storage means in response to a search request input from the user terminal, and evaluates the document. From the search behavior in the user terminal to perform, obtain an evaluation for the document, calculate an evaluation value for each document based on the obtained evaluation,
An information search program for ordering the output of a search result in response to a search request from a user terminal using the evaluation value, wherein the search request, the information specifying the document, and the evaluation of the document are evaluated based on search behavior in the user terminal. The search request, the document identification information, and the evaluation of the document are stored in the document evaluation log storage unit, the direct evaluation of the document directly evaluated at the user terminal is calculated, and the search request and the search request from the user terminal are obtained. The indirect evaluation of the document is calculated using the information of the evaluation of the document obtained from the user terminal for the document that is the search result for the search request, and the indirect evaluation of each document is calculated from the calculated direct evaluation and indirect evaluation of each document. The gist is to calculate a comprehensive evaluation value and store the comprehensive evaluation value in the document evaluation storage means corresponding to each document.

According to the thirteenth aspect of the present invention, a search request, document identification information, and an evaluation of a document are obtained from a search action at a user terminal, and a direct evaluation of the document directly evaluated at the user terminal is calculated. Then, for a search request from the user terminal and for a document that is a search result for the search request, an indirect evaluation of the document is calculated using information of the evaluation of the document obtained from the user terminal. Calculates the comprehensive evaluation value of each document from the direct evaluation and indirect evaluation of the document, and stores the comprehensive evaluation value corresponding to each document. The indirect evaluation can be performed from the viewpoint of the evaluation, and a search request given from a user terminal to each document for a group of a plurality of documents and a past user terminal can be obtained. When the user terminal can rank and output the documents requested by the user terminal from the collection of documents using the evaluation value given by using these evaluations, and there are not enough evaluations to perform accurate evaluation on the documents If there is a document that does not have an evaluation such as a new document or a new document, use the existing evaluation and the user's search log to identify the part of the directly evaluated document that would have been evaluated by the user. And the number of document evaluations can be increased without lowering the quality of the document evaluation.

According to a fourteenth aspect of the present invention, in the invention according to the thirteenth aspect, the processing for calculating the indirect evaluation is performed based on a search request from the user terminal and a document that is a search result corresponding to the search request. Using the information of the evaluation of the document obtained from the above, obtain a partial document of interest at the user terminal from the directly evaluated document, which is a document directly evaluated at the user terminal, and The gist of the present invention is to determine a high document and obtain an indirect evaluation of the document from the directly evaluated document from the direct evaluation value of the directly evaluated document and the relevance of a certain document to the directly evaluated document.

According to the fourteenth aspect of the present invention, a partial document that is noticed by a user terminal is obtained from a directly evaluated document by using a search request and information on document evaluation, and the relevance of the partial document to the partial document is obtained. In order to determine the high document and obtain the indirect evaluation of the document from the directly evaluated document from the direct evaluation value of the directly evaluated document and the relevance of a certain document to the directly evaluated document, an accurate When there are not enough evaluations to perform the evaluation, or when there is a document that does not have an evaluation such as a new document, the existing evaluation and the user's search log are used to evaluate the actual Thus, it is possible to identify a portion that would have been evaluated by the user, and to increase the number of document evaluations without deteriorating the quality of the document evaluation.

According to a fifteenth aspect of the present invention, a document obtained by searching a set of a plurality of documents from a document storage unit in response to a search request input from a user terminal is referred to in the user terminal to evaluate the document. From the search behavior in the user terminal to perform, obtain an evaluation for the document, calculate an evaluation value for each document based on the obtained evaluation,
A recording medium recording an information search program for ordering the output of search results in response to a search request from a user terminal using the evaluation value, wherein the search request, information specifying the document, and evaluation of the document are performed by a user. Obtained from the search behavior at the terminal, store the search request, the document identification information, and the evaluation for the document in the document evaluation log storage unit, calculate the direct evaluation for the document directly evaluated at the user terminal, The indirect evaluation of the document is calculated using the information of the evaluation of the document obtained from the user terminal for the search request of the document and the document that is the search result corresponding to the search request, and the calculated direct evaluation and indirect of each document are calculated. An information retrieval program for calculating a comprehensive evaluation value of each document from the evaluation and storing the comprehensive evaluation value in the document evaluation storage means corresponding to each document. And summarized in that to record the ram on the recording medium.

According to the fifteenth aspect of the present invention, the search request, the document specifying information, and the evaluation of the document are obtained from the search behavior at the user terminal, and the direct evaluation of the document directly evaluated at the user terminal is calculated. Then, for a search request from the user terminal and for a document that is a search result for the search request, an indirect evaluation of the document is calculated using information of the evaluation of the document obtained from the user terminal. The total evaluation value of each document is calculated from the direct evaluation and the indirect evaluation of the document, and the information retrieval program that stores the overall evaluation value corresponding to each document is recorded on a recording medium. Its distribution can be improved.

According to a sixteenth aspect of the present invention, in the invention according to the fifteenth aspect, the processing for calculating the indirect evaluation is performed on the basis of a search request from the user terminal and a document that is a search result corresponding to the search request. Using the information of the evaluation of the document obtained from the above, obtain a partial document of interest at the user terminal from the directly evaluated document, which is a document directly evaluated at the user terminal, and A recording medium for determining a high document and obtaining an indirect evaluation of the document from the directly evaluated document based on the direct evaluation value of the directly evaluated document and the relevance of a certain document to the directly evaluated document. It shall be recorded in the summary.

According to the sixteenth aspect of the present invention, a partial document that is noticed by a user terminal is obtained from a directly evaluated document by using a search request and information on document evaluation, and the relevance of the partial document to the partial document is obtained. An information search program for determining a high document and obtaining an indirect evaluation of the document from the directly evaluated document based on the direct evaluation value of the directly evaluated document and the relevance of a certain document to the directly evaluated document is recorded on a recording medium. Therefore, the distribution of the recording medium can be improved by using the recording medium.

[0045]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a configuration of a server system to which the information search device according to the first embodiment of the present invention is applied. Server system 110 shown in FIG.
Is a search system using the WWW. A client system 151, which is a plurality of user terminals, is connected via a network such as an interface, and these client systems 151 have a built-in browser 153.

The server system 110 is connected to the client system 151 via a network, receives a search request from the client system 151 via the browser 153 to search for a user, and responds to the search request as shown in FIG. A search form transmission unit 111 for returning such a search form to the browser 153 is provided.

As shown in FIG. 2, the search form includes a search condition column for inputting a search condition including a search word such as a keyword, a list of search viewpoints for selecting according to the purpose of use of information, That is, in FIG. 2, for example, a list of search viewpoints such as "I want to know new information", "I want information on functions, products, events, etc.", "I want to know prices", "I want to know what means for realization" are displayed. In accordance with the search viewpoint list display field that is being displayed, the search viewpoint display field that displays the search viewpoint that is the search purpose selected from this list, and the selected search viewpoint and the search condition that is input in the search condition column. An execution key for executing a search is displayed, and the user of the client system 151 inputs desired search conditions and search viewpoints in the respective columns while viewing the search form, and executes the search. The search condition and search the viewpoint is sent to the server system 110 as a search request by clicking or the over, so that the search is performed. The list of search viewpoints is stored in the server system 11.
On the side of managing 0, typical and frequently used search viewpoints are obtained and listed from the system user and documents to be searched.

When the user inputs data while viewing the search form in the client system 151 and further executes a search, the search request transmitted from the browser 153 of the client system 151 and including the search conditions and the search viewpoint is transmitted via the network. Server system 110
Is received by the search request receiving unit 113. The search request receiving unit 113 passes this search request to the search management unit 115.

Upon receiving the search request including the search conditions and the search viewpoint, the search control unit 115 transfers the search conditions to the document search unit 117. The document search unit 117 searches the document database 119 based on the search condition, and obtains a document ID group that meets the search condition. The obtained document ID is returned to the search control unit 115. The document database 119 stores a document ID for specifying each document in correspondence with a keyword that is a word included in each document.

The search control unit 115 passes the document ID from the document search unit 117 and the search viewpoint to the document evaluation acquisition unit 121. The document evaluation obtaining unit 121 searches the document evaluation database 123 based on the document ID and the search viewpoint, obtains each evaluation value of the document ID in the search viewpoint, and obtains the document ID and the score of the evaluation value. The combination is passed to the search control unit 115.

As shown in FIG. 8, the document evaluation database 123 divides an evaluation value, which is the relevance of each document with respect to each search viewpoint, that is, a score, for each search viewpoint and for each document ID, and for each period. Stored.

The search control unit 115 is the document evaluation acquisition unit 12
Receiving a combination of each document ID and a score that is its evaluation value from 1, the document IDs are arranged in the order of the scores,
From the document IDs with the highest scores, the specified number of documents are created as search results, and the search results are sent to the search result transmitting unit 1.
25 through the network to the client system 1
51 to the browser 153.

As described above, as a result of the user browsing the search result in the client system 151 to which the search result has been returned by the search result transmitting unit 125, the behavior information including the user's browsing or evaluation of the search result is the target of the behavior. Is transmitted from the browser 153 of the client system 151 to the server system 110 via the network, together with the document ID and the search viewpoint, and the search behavior receiving unit 127 receives the document ID.

More specifically, the search behavior receiving section 127 outputs information indicating that the user has referred to one document at the time of selecting one document on the search result display screen as shown in FIG. 3, or as shown in FIG. At the time when the “evaluation” of the selected document on the document display screen is performed, the information that “evaluated” is received together with the search point and the document ID initially specified, and
Deliver to the search behavior registration unit 131. If the information indicates that the document has been “referenced”, the result document transmitting unit 129
Hand over the information as well. Upon receiving the information, the result document transmitting unit 129 transmits a document display screen as shown in FIG. 4 to the browser 153 as a search result based on the information, that is, based on the information referred to by the user, and displays the screen. You.

As shown in FIG. 3, a search result display screen presents search results created based on a user's search request. The hyperlink in this screen does not directly specify the target document, but specifies that the document is displayed via the system. As a result, it is possible to acquire "reference" of the user's search behavior, and to display a search result document in the form shown in FIG.

On the document display screen shown in FIG. 4, a screen of a search result document is displayed, and "evaluation" of this document is received. In addition, a menu column is provided in which a search viewpoint can be selected again to perform “evaluation” from a viewpoint different from the search viewpoint input at the time of the search.

The search action registration section 131 stores the document ID, user action information such as reference or evaluation, and the search viewpoint received and received from the search action reception section 127 in the search action database 133 together with the reception time.

As shown in FIG. 7, the search action database 133 stores action information such as document ID, date and time, search viewpoint, user reference or evaluation for each user action. .

Each piece of information stored in the search action database 133 is monitored by the search action analysis section 135, and the search action analysis section 135 stores a predetermined amount or more in the search action database 133 every time a predetermined time elapses. When data is accumulated, the search behavior database 13
3, the user's operation on the document in a predetermined period is analyzed based on the information, and the evaluation value of the document in the search viewpoint is calculated based on the analysis result. And the score which is the evaluation value for each document ID is updated.

In the analysis processing in the search behavior analysis unit 135, the related document evaluation unit 137 acquires a document highly relevant to the document evaluated by the user from the information in the document database 119 and the search behavior database 133, By passing the information of the related document to the search behavior analysis unit 135, even for a document that has not been directly evaluated by the user, a score that is an evaluation value is calculated using the evaluation value of the document related to the document. Update. That is,
The search behavior analysis unit 135 performs the search behavior database 1
33, the related document evaluation unit 1
The user's operation on the document is analyzed based on the information of the related document received from 37, and a score that is the evaluation value of the document in the search viewpoint is calculated based on the analysis result. And the score for each document ID is updated.

Next, the operation of the information retrieval apparatus configured as described above will be described with reference to the flowcharts shown in FIGS. The operation of the information retrieval device is as follows.
It is divided into two phases: an information retrieval phase and a construction phase of a document evaluation database. First, the information search phase will be described with reference to the flowchart shown in FIG.

First, the search form transmitting unit 1 of the server system 110 issues a search request from the client system 151 via the browser 153 to search for a user.
11 is accepted, the search form transmitting unit 111 transmits the search form shown in FIG. 2 to the browser 153 (step S11). In accordance with the search form, the user creates a search request by inputting search conditions and specifying a search viewpoint (step S12). Browser 153
Transmits the search form input with the search request to the server system 110 (step S13).

The search request transmitted from the browser 153 is received by the search request receiving unit 113, and the search control unit 11
(Step S15). Search Management Department 11
In step 5, a search condition and a search viewpoint are extracted from the delivered search request (step S17). The search control unit 115 transfers the search condition to the document search unit 117, and acquires the ID of the document that includes or is similar to the search condition from the document database 119 (step S19).

The search control unit 115 transmits the search viewpoint and the acquired document ID to the document evaluation acquisition unit 121, and acquires the score of each document from the specified search viewpoint (step S21). The search control unit 115 sorts the obtained document IDs in the order of the scores, obtains a predetermined number of results from those having a high degree of importance, and creates a search result (step S23). Note that the number of search results displayed, the order in which search results are displayed, and the like can be made variable by designating the search request.

The search result is displayed as a search result on the browser 153 of the client system 151 via the search result transmitting unit 125 (step S25). This screen is shown in FIG. Information that the user has selected one document from the search result display screen displayed on the browser is transmitted from the search behavior receiving unit 127 to the result document transmitting unit 129,
The document requested by the user is displayed on the browser 153 (steps S27-S33). This document display screen is shown in FIG.

The information that this selection is made is made by the document “reference”.
The information is received from the search behavior receiving unit 127 and passed to the search behavior registration unit 131. At the same time, the search viewpoint specified at the time of the search and the ID of the target document are also delivered (step S27). Thereafter, the process proceeds to the "document evaluation database" construction phase (step S29).

The user uses the button for evaluating a document arranged at the bottom of the document display screen to perform "evaluation" of the document.
Can be entered. In addition, a list of the same viewpoint as that at the time of the search is presented, and by selecting, a user can perform evaluation from a viewpoint different from the currently selected viewpoint while browsing the document.

If the search result is directly evaluated (step S35), the information indicating that the evaluation has been made is transferred from the search action receiving section 127 to the search action registration section 131 (step S36). At the same time, the search viewpoint specified at the time of the search and the ID of the target document are also delivered. Thereafter, the process proceeds to the "document evaluation database" construction phase (step S37). Then, the search result screen is displayed again or the current search is ended by the user's operation (steps S39 and S41).

Next, the construction phase of the document evaluation database 123 used for presenting search results based on the search viewpoint in this system will be described with reference to the flowchart shown in FIG.

As described above, the search behavior receiving section 127
Transfers the information of “reference” or “evaluation”, the document ID, and the search viewpoint to the search behavior registration unit 131 (step S6).
1). The search action registration unit 131 adds date and time information to the transferred information, and registers the information in the search action database 133 (step S63).

At regular intervals or when a predetermined amount or more of data is stored in the search behavior database 133, the search behavior analysis unit 135 causes the search behavior database 13
The evaluation of the document for each search viewpoint is calculated using the information in No. 3 and registered in the document evaluation database 123 (step S65).

Further, by using the related document evaluation unit 137, based on the information in the document database 119 and the retrieval behavior database 133, it is possible to input the evaluation for the document evaluated by the user and the highly relevant document. Is also possible.

Next, as an application example of the present embodiment, the behavior of the user when applied to an actual system will be described.

First, the user inputs a search keyword (search word) as a search condition and selects a search purpose from a list as a search viewpoint from an interface which is a search form shown in FIG. 2, and presses an execution button. As a result, the user obtains the search result shown in FIG.

From this search result, the user refers to the information of each document including the summary, title, etc., and determines the document to be actually referred to. When the document is determined, the user clicks the link of the target document to “reference” the document. As a result, the user obtains the search result screen shown in FIG.

After the user refers to the displayed document, the user, for example, “good”, “ba”
"Evaluation" can be input with an evaluation input button such as "d". Thus, the user obtains one document. If the document is not what is desired or if more information is desired, a new document can be obtained by repeating the above process.

Next, the construction of the document evaluation database 123 will be described. As described above, the information of “reference” or “evaluation” input by the user is written in the search behavior database 133 as one entry of the search viewpoint, the ID of the target document, and the date and time.

At a stage where these entries have gathered to some extent or at a stage where some time has passed, analysis of the search behavior database 133 and construction of the document evaluation database 123 are performed. In other words, the construction of the document evaluation database 123 is basically to add a score of a document from a certain viewpoint when the document has been evaluated well.
Then, a higher score is assigned to a document that has received a better evaluation, and the score is reduced in the case of a bad evaluation.

Thus, the document evaluation database 1
When a new search is performed as a result of the creation of the search result 23, the search result is presented at a higher position from the document that has obtained a higher score from the search viewpoint specified by the user among the documents in the document set squeezed by the search keyword. . This allows the user to preferentially receive presentations with higher evaluations. In addition, the evaluation is always received, and the effect of the old evaluation is attenuated or eliminated so that the evaluation corresponding to the time can be obtained.

Next, a method for creating and updating the document evaluation database 123 by the search behavior analysis unit 135 and the search and analysis of the document evaluation database 123 for each search viewpoint and document ID,
A method of calculating an evaluation value, which is a score that is registered for each period, will be described in detail.

The search behavior analysis unit 135 first divides each record in the search behavior database 133 into time periods using date and time information, and classifies the records in each period into each search viewpoint.

Then, the set of records for each search viewpoint is classified by document ID, and the records are classified for each search viewpoint and document ID. Then, the types of evaluation (reference, evaluation (good evaluation, bad evaluation, etc.)) are classified and acquired from one classified record. An evaluation value, that is, a score is calculated from time information based on the period classified as described above, the number of references, and the number of evaluations, and is stored in the document evaluation database 123 as shown in FIG.

Next, an information retrieval apparatus according to a second embodiment of the present invention will be described.

In the second embodiment, as described above,
In a method that uses the direct evaluation of the user, reference is performed unconsciously by the user by acquiring a document that is performed at the time of retrieval, whereas a sufficient number of direct evaluations are generally obtained because the input load of the user is large. No, but in spite of the small number of ratings themselves, in addition to trying to properly rank search results using ratings for documents,
When ranking documents using the user's past search history, the new document has no past evaluation, so there is no evaluation value, and the new document is ranked lower in the search results,
In order to solve the problem that a new document is presented to the user too little and is buried, the document is evaluated even if there is no direct evaluation from the user. In addition to the user's direct evaluation, a document highly relevant to the document evaluated by the user is to be evaluated according to the evaluation.

However, when the relevance between documents is evaluated based on the keyword frequency between documents in a similar document search that is generally performed, especially when a plurality of topics exist in the document, In this embodiment, information obtained from the user as a document search device, that is, a user search request (which may include a search keyword and other information),
The relevance of the document is evaluated from the viewpoint of the user's search by combining information specifying the target document such as the document ID to be evaluated and information including the contents of the evaluation of the document, and using this information. Thus, indirect evaluation is performed with high accuracy on a document highly relevant to the document evaluated by the user.

Specifically, in the direct evaluation, an evaluation value is calculated from a search history of a user who has directly evaluated each document. The indirect evaluation includes (1) acquisition of a partial document considered to have been evaluated by the user, (2) acquisition of a document highly related to the acquired partial document, and (3) evaluation of the user to the directly evaluated document. The indirect evaluation value for the related document is calculated using the value and the relevance value.

In the acquisition of the partial document considered to have been evaluated by the user (1), a partial document is acquired from the target document by using the search keyword as a clue, and thereby the user evaluates the partial document. Get the partial document that would have been referenced.

In the acquisition of a document having a high degree of relevance to the acquired partial document (2), the frequency and distribution of keywords existing in the partial document are analyzed, and the partial document considered to have been evaluated by the user is obtained. Identify highly relevant documents and acquire them with their relevance.

Further, in the calculation (3) of the indirect evaluation value for the related document using the evaluation value of the user and the value of the relevance of the document to be directly evaluated, the indirect evaluation value is calculated using the relevance. Accordingly, the degree of indirect evaluation can be changed depending on the degree of relevance of the document.

By determining the overall evaluation value of a document from the values of the direct evaluation and the indirect evaluation described above, it is possible to give an evaluation value to a document that is not directly evaluated.

Next, the present embodiment will be described with reference to the drawings. FIG. 9 is a block diagram showing a configuration of a document search device to which the information search device of the second embodiment is applied.
In the document search apparatus shown in FIG. 2, a plurality of user terminals, which are client systems, are connected via a network via communication means with the client system as in FIG.

The document search apparatus shown in FIG. 9 includes a search unit 11 forming a document search unit, a document database 12 forming a document storage unit, a document evaluation database 13 forming a document evaluation storage unit, and a document evaluation obtaining unit. A document evaluation log database 15 forming a document evaluation log storage unit, and a document evaluation unit 20 forming a document evaluation unit. In this embodiment, it is assumed that the search request is composed of only the search keyword, and that the document has only one comprehensive evaluation value for each document. Example 1
It is also possible to add additional information other than the search request to the search request as in the search viewpoint shown in FIG.

The document database 12 stores what words exist in each document to be searched and in what order. The document evaluation database 13 stores an overall evaluation value of each document.

The search unit 11 receives a search keyword from the user terminal, searches the document database 12 with the received search keyword, and acquires a document set including the search keyword. Then, the user accesses the document evaluation database 13, acquires the comprehensive evaluation value of each document in the acquired document group from the document evaluation database 13, ranks each document based on the comprehensive evaluation value, and obtains the search result. Create and return to user terminal.

The document evaluation obtaining unit 14 obtains the evaluation of the document from the user terminal, and stores three pieces of information including the search keyword, the directly evaluated document ID, and the evaluation of the document in the document evaluation log database 15 as one record. I do.

The document evaluation log database 15 stores, as one record, three pieces of information including a search keyword, a directly evaluated document ID, and a document evaluation. When the search request includes the search viewpoint, as shown in FIG. 12, in addition to the document ID that is the directly evaluated document ID, the keyword that is the search keyword, and the evaluation of the document, the viewpoint that is the search viewpoint is also stored. Can be.

As shown in detail in FIG. 10, the document evaluation section 20 includes an evaluation management section 21, a direct evaluation section 22, a temporary storage section 24, and an indirect evaluation section 30.

The evaluation management unit 21 obtains a log stored in the document evaluation log database 15 as shown in FIG. 12, and stores the obtained log as an entry for each directly evaluated document ID as shown in FIG. Divide into sets. then,
In this way, the divided entry set for each directly evaluated document ID is sequentially supplied to the direct evaluation unit 22 to obtain the direct evaluation. If the search request includes a search viewpoint, the logs are totaled in the form of FIG. 13 for each viewpoint.

Next, the entry for each of the directly evaluated document IDs is passed to the indirect evaluation unit 30 to obtain the indirect evaluation value of each document by each of the directly evaluated documents. These direct evaluation values and indirect evaluation values are temporarily stored in the temporary storage unit 24. When the search request includes a search viewpoint, the temporary storage unit 24 stores the data in a data format for each viewpoint as shown in FIG.

After the evaluation of all logs of the directly evaluated document IDs is completed, the indirect evaluation value of each document is calculated, the total evaluation value is calculated from the direct evaluation value and the indirect evaluation value, and the document evaluation database is calculated. 13 is stored.

The direct evaluation unit 22 obtains a log of each directly evaluated document, calculates a direct evaluation value, and returns the direct evaluation value to the evaluation management unit 21.

As shown in detail in FIG. 11, the indirect evaluation unit 30 comprises an indirect evaluation management unit 31, a partial document acquisition unit 32, a document relevance determination unit 33, and a temporary storage unit.

The indirect evaluation management unit 31 acquires a log entry set of each directly evaluated document ID, and further divides the log entry set for each keyword as shown in FIG. From the log divided for each keyword, a basic evaluation value is calculated based on a calculation formula similar to the calculation of the direct evaluation value, and the directly evaluated document I is calculated.
D and the search keyword are supplied to the partial document acquisition unit 32, and a partial document that is considered to have been focused on by the user when performing the evaluation is acquired. Then, the obtained partial document is supplied to the document relevance determination unit 33, the value of the ID of the relevant document and the value of the relevancy are obtained, and the indirect evaluation value for each keyword is calculated from the relevance and the basic evaluation value.

The indirect evaluation value for each keyword is shown in FIG.
Are temporarily stored in the temporary storage unit 34 in a data format as shown in FIG. By performing this for all keywords, the indirect evaluation value of each document for one directly evaluated document is obtained and returned to the evaluation management unit 21. When the search request includes the search viewpoint, as shown in FIG.
Stored for each viewpoint.

The partial document acquisition unit 32 is an indirect evaluation management unit 3
Using the document ID and the search keyword received from 1, a partial document that is considered to be of interest when the user evaluates the document is obtained from the document database 12 and supplied to the document relevance determining unit 33.

The document relevance judging unit 33, based on the partial document received from the indirect evaluation management unit 31 and the data of the document database 12, identifies the ID of the document considered to be related to the partial document, the partial document, and the The association is returned to the indirect evaluation management unit 31 as a set.

Next, an operation procedure when the document evaluation database 13 is created in the present embodiment will be described with reference to a flowchart shown in FIG. When the document evaluation database 13 is created, it is assumed that the document evaluation log database 15 stores a document evaluation (reference and direct evaluation) log by the user of the document search device.

In FIG. 17, first, the evaluation management unit 21 of the document evaluation unit 20 acquires entries of the document evaluation log database 15 and compiles them for each directly evaluated document ID (step S71). The evaluation management unit 21 receives the evaluation document I directly.
The set of entries compiled for each D is sequentially evaluated directly by the evaluator 22.
To supply. The direct evaluation unit 22 calculates an evaluation value directly from the entry set and returns it to the evaluation management unit 21 (Step S73).

The evaluation management unit 21 stores the ID of the directly evaluated document.
The entry set compiled for each is sequentially sent to the indirect evaluation unit 30 (step S75). In the indirect evaluation unit 30, the indirect evaluation management unit 31 divides the entry set of one directly evaluated document ID for each search keyword (step S7).
7). The indirect evaluation management unit 31 calculates an evaluation value for each keyword from the divided entry group by the same method as the direct evaluation (step S79). This evaluation value is called a basic evaluation value.

The indirect evaluation management unit 31 is configured to acquire the partial document
The document ID to be directly evaluated and the search keyword are transmitted to 2 to obtain a partial document (step S81). And
The indirect evaluation management unit 31 supplies the acquired partial document to the document relevance determination unit 33, and outputs the document ID associated with the partial document.
And the relevance with the partial document (step S8)
3).

The indirect evaluation management unit 31 calculates an indirect evaluation value based on the relevance and the basic evaluation value, and calculates an indirect evaluation value when the directly evaluated document is searched and evaluated by the search keyword (step S1). S85). Then, the indirect evaluation management unit 31 determines whether or not the above-described processing has been completed for all the search keywords while storing the indirect evaluation value in the temporary storage unit 34 (step S87). , The process returns to step S79, and the same processing as described above is performed for all search keywords.

When the search is completed for all the search keywords, the indirect evaluation management unit 31 sums up the indirect evaluation values obtained from the case of each keyword stored in the temporary storage unit 34 in each document, and The indirect evaluation value of each document from the directly evaluated document, and I
D, the document ID, and the indirect evaluation value are transmitted to the evaluation management unit 21. That is, the indirect evaluation value when the directly evaluated document of each document is searched and evaluated by the search keyword is calculated based on the indirect evaluation value by the search keyword, and transmitted to the evaluation management unit 21 (step S89). .

Then, the evaluation management section 21 stores the temporary storage section 2
While storing the direct evaluation value and the indirect evaluation value in No. 4, it is determined whether or not the above processing has been performed for all the directly evaluated documents (step S91), and when all the directly evaluated documents have not been processed. In step S73
And the same process is repeated for all the directly evaluated documents.

When all of the documents to be directly evaluated have been completed, the evaluation management unit 21 performs a comprehensive evaluation of each document based on the direct evaluation and the indirect evaluation stored in the temporary storage unit 24 for each document of each document ID. The value is calculated and registered in the document evaluation database 13 (step S93).

As described above, in the present embodiment, it is possible to perform indirect evaluation from the viewpoint of performing direct evaluation at the user terminal for a document not directly receiving evaluation from the user terminal. Using the search request given by the user terminal and the evaluation value given from the past user terminal to each document collection, rank the documents that the user terminal wants from the document collection using the evaluation value given using the past user terminal evaluation When there are not enough evaluations to be able to output and perform accurate evaluation on the document, or when there is a document that does not have an evaluation such as a new document, the existing evaluation and the user's search log are used. It is possible to specify a portion of the directly evaluated document that would have been actually evaluated by the user, thereby increasing the number of evaluations of the document without deteriorating the quality of the document evaluation.

Next, a method of calculating the document evaluation (comprehensive evaluation) will be described. First, in accordance with the first embodiment, in order to obtain a more accurate evaluation of the user, the user refers to the document in the search result as a degree of popularity. Is performed by using "direct evaluation", which performs direct evaluation. Calculate using these,
The evaluation based on the popularity of each document for each “search purpose” is referred to as “overall evaluation”. However, if the evaluation of a document, that is, the “comprehensive evaluation” is calculated by simply adding the number of “reference” and “direct evaluation”, the following points become problems.

The first problem is that “reference” and “direct evaluation” are concentrated on documents having a high ranking. In general, documents having a higher ranking in the search ranking are frequently seen by the user, and are easily referred to. As a result, the evaluation concentrates on the high ranking documents, the evaluation does not reach the entire document set, and the search results are biased.

The second problem is that the evaluation becomes obsolete. That is, stale documents continue to maintain a high ranking based on past evaluations.

As a solution to the first problem, for the document i, “reference” and “direct evaluation” in the “search purpose” j are calculated by the following method, thereby eliminating the ranking-dependent element.

"Reference value" R _ij : The frequency of the number of references greatly depends on the number of pages displayed in the search results of the document. Therefore, the reference value performed on the subsequent page is larger than that on the first page. That is, when a document is referred to as “reference value”, “the number of references when the document exists on one page”
Is used. The formula for calculating the “reference value” when the reference value correction function is F (r) (r is the number of pages of the search result in which the document appears) is as follows.

[0121]

(Equation 1) Here, F (r) is used to investigate the behavior of the user in the conventional search system (how often the search result page is referred to and how many pages are referred) with respect to the number of pages. Is a value indicating how rare the document is with respect to the document on the first page. This time, using the log in the past information retrieval system, the frequency distribution was approximated by the least squares method, and the result was as follows.

F (r) = r ^2.08 (2) “Direct evaluation value” E _ij : The number of “direct evaluations” does not depend on the direct ranking, but clearly increases or decreases with the increase or decrease of the number of “references”. Conceivable. Therefore, as the “direct evaluation value”, “the number of“ direct evaluation ”times when“ reference ”is performed on each document on average” is used. The "average number of references" is calculated from the number of combinations of all "referenced" documents and "search purpose" in the evaluation log and the total number of "references" to the relative value to the number of "references". By obtaining the product of, "direct evaluation value" is obtained.

Although it is possible to fairly evaluate a document referred to by the method described above, there is no difference that a document not referred to always exists in the lower rank of the ranking. Therefore, in the present embodiment, "negative evaluation" is used for "direct evaluation" in order to positively lower documents that do not match the selected "search purpose" while being at the top of the ranking. This aims to push documents that are relatively lower in the ranking to higher rankings, and to reduce documents that do not have “direct evaluation”. g _ij , b _ij
Is the number of “plus evaluations” and “minus evaluations” obtained in the logs, N is the number of combined documents and objectives that were “referenced” between the previous update and this update, n is the total number of documents, and m is the total number of documents. Assuming that the total number of “search purposes” is “direct evaluation value”, it is as follows.

[0124]

(Equation 2) Next, indirect evaluation values performed in addition to the direct evaluation as described in the second embodiment will be described.

Here, each symbol used for calculating the indirect evaluation value is defined as follows.

Indirect evaluation value of document i for “search purpose” j: E _ij ^I Indirect evaluation of document i for document k: Document set that is relevant to E _ik i and directly evaluated: C _i search request ( Search keyword): q Q _k is a set of search requests in which k has been searched For search request q, relevance of document i to document k:
"Search purpose" j when the search is performed by the search request q of the _Rikq document k
E _kqj ^{D The} indirect evaluation value E _ij ^I is defined as follows:

[0127]

(Equation 3) Further, the direct evaluation value E _ij defined by the above equation (3)
Is replaced by E _ij ^D , the evaluation value E _ij is finally defined by the following equation.

[0128]

(Equation 4) Here, γ is a weight given to the indirect evaluation value, and γ <0 from the viewpoint of emphasizing direct evaluation.

Next, as a solution to the second problem, a time parameter is added to the evaluation function. By adding a time parameter to the overall evaluation, a difference is made between the past evaluation and the latest evaluation. As a result, a document that has been evaluated only for one period loses its evaluation over time, and its ranking is gradually lowered as a document of low importance. Conversely, documents that are always evaluated maintain a high ranking. The calculation of the comprehensive evaluation was performed at regular intervals, and the evaluation was attenuated by 10% each time the evaluation was updated. Here, if the freshness of the information t times before is T (t), the following equation is obtained.

[0130]

(Equation 5) Next, the calculation of the overall evaluation will be described. In order to use the “evaluation value” and the “reference value” as values of the same level in calculating the overall evaluation, they are expressed as values when the number of “reference” times and the number of “direct evaluation” times are the same. For this purpose, the “evaluation value” is corrected by the following coefficient: α.

[0131]

(Equation 6) In addition, since “direct evaluation” which is originally a direct opinion of the user is considered to be more important than “reference”, “evaluation value” is weighted. The weight β is determined using the ratio of the sum of the total number of “references” and the total number of “direct evaluations” actually obtained for the weighting. That is, weighting is performed according to each frequency ratio.

[0132]

(Equation 7) By the above method, the “overall evaluation” S _ij for the “search purpose” j of the document i is determined by the following equation. Here, 1 is the total number of updates, and R _ij ^t and E _ij ^t are R _ij and E _ij obtained from the log t times before, respectively.

[0133]

(Equation 8) The processing procedure of the information search method of the above embodiment is recorded as a program on a recording medium such as a CD or FD, and the program recorded on the recording medium is downloaded to a computer system via a communication line, or By installing the program from a recording medium and operating the computer system with the program, it is possible to function as an information retrieval apparatus that implements the information retrieval method. Can be enhanced.

[0134]

As described above, according to the present invention,
When a search request including a search condition including a keyword for limiting a search range and a search viewpoint based on a user's search purpose is received as a search request, a document database is searched based on the search condition, and a document ID matching the search condition is obtained. The document evaluation database is searched based on the document ID and the search viewpoint, the evaluation values of the document IDs in the search viewpoint are obtained, the document IDs are arranged in the order of the evaluation values, and the document IDs having the higher evaluation values are designated. The search results are created as search results and returned to the user terminal, so that the search viewpoint can be added to the search request so that the search request can be transmitted more clearly. Can be removed and output accurate search results that match the search request,
The user can easily reach the target document, and can present the user with a search result based on the evaluation of the user who has performed a search from the same search viewpoint in the past. In addition, it is possible to acquire data indicating a user's preference from each viewpoint by an information collection mechanism such as a document evaluation database, thereby providing a better service to an individual or the entire user. .

Further, according to the present invention, as a result of browsing a search result on a user terminal, behavior information including browsing or evaluation of a user is received, stored in a search behavior database, and stored in the search behavior database. Analyzes the user's operation based on the information, calculates the evaluation value of the document in the search viewpoint based on the analysis result, and updates the evaluation value of the document evaluation database with the evaluation value. Therefore, the evaluation value stored in the document evaluation database Will always reflect the user's actions, and the search results retrieved from this document evaluation database will always reflect the user's browsing and evaluation actions, and present the user with the optimal search results that match the user's search purpose. can do. Furthermore, according to the present invention,
It is possible to perform indirect evaluation from the viewpoint of performing direct evaluation at the user terminal for documents that have not been directly evaluated by the user terminal, and to give a document set from the user terminal to a set of multiple documents It is possible to rank and output the documents requested by the user terminal from a collection of documents using the evaluation value given by using the search request given and the evaluation from the past user terminals, and to perform accurate evaluation on the documents When there are not enough evaluations in a document or when there is a document that does not have an evaluation such as a new document, the user actually evaluates the document to be evaluated directly using the existing evaluation and user search logs. It is possible to specify a portion that may have been evaluated, and it is possible to increase the number of evaluations of the document without deteriorating the quality of the document evaluation.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an information search device according to a first embodiment of the present invention.

FIG. 2 is a diagram showing a search form displayed on a browser of a client system as a user terminal in the information search device of the first embodiment shown in FIG.

FIG. 3 is a diagram showing a search result display screen displayed on a browser of a client system which is a user terminal in the information search device of the first embodiment shown in FIG. 1;

FIG. 4 is a diagram showing a document display screen displayed on a browser of a client system as a user terminal in the information search device of the first embodiment shown in FIG.

FIG. 5 is a flowchart showing an operation of an information search phase of the information search device according to the first embodiment shown in FIG. 1;

FIG. 6 is a flowchart showing an operation of a document evaluation database construction phase of the information search device of the first embodiment shown in FIG. 1;

FIG. 7 is a diagram showing a configuration of a search behavior database used in the information search device of the first embodiment shown in FIG.

FIG. 8 is a diagram showing a configuration of a document evaluation database used in the information search device of the first embodiment shown in FIG.

FIG. 9 is a block diagram illustrating a configuration of a document search device to which the information search device according to the second embodiment of the present invention is applied.

FIG. 10 is a block diagram showing a detailed configuration of a document evaluation unit used in the document search device according to the second embodiment shown in FIG.

FIG. 11 is a block diagram showing a detailed configuration of an indirect evaluation unit used in the document search device according to the second embodiment shown in FIG.

FIG. 12 is a diagram showing a configuration of a document evaluation log database used in the document search device of the second embodiment shown in FIG.

FIG. 13 is a diagram showing a state where a log stored in a document evaluation log database used in the document search device of the second embodiment shown in FIG. 9 is divided into an entry set for each directly evaluated document ID. It is.

FIG. 14 is a diagram showing a state in which an entry set divided for each directly evaluated document ID as shown in FIG. 13 is divided for each keyword;

FIG. 15 is a diagram showing a data format for each viewpoint stored in a temporary storage unit of the document evaluation unit shown in FIG.

16 is a diagram showing a data format of an indirect evaluation value for each keyword stored in a temporary storage unit of the indirect evaluation unit shown in FIG.

17 is a flowchart showing an operation procedure when creating a document evaluation database in the embodiment of the document search device of the second embodiment shown in FIG.

[Explanation of symbols]

 Reference Signs List 11 retrieval unit 12, 119 document database 13, 123 document evaluation database 14, 121 document evaluation acquisition unit 15 document evaluation log database 20 document evaluation unit 21 evaluation management unit 22 direct evaluation unit 30 indirect evaluation unit 31 indirect evaluation management unit 32 partial document Acquisition unit 33 Document relevance determination unit 110 Server system 111 Search form transmission unit 113 Search request reception unit 115 Search supervision unit 117 Document search unit 125 Search result transmission unit 127 Search behavior reception unit 129 Result document transmission unit 131 Search behavior registration unit 133 Search behavior database 135 Search behavior analysis unit 137 Related document evaluation unit 151 Client system 153 Browser

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5B075 KK07 ND16 NK02 PP02 PP22 PQ02 PQ42 PQ46 PR04 PR08 QM08 UU40

Claims (16)

[Claims] 1. An information retrieval apparatus for retrieving and outputting information conforming to a search request in response to a search request input from a user terminal, wherein the search condition includes a keyword for limiting a search range as the search request. And a search request receiving means for receiving a search viewpoint for evaluating the priority of information based on a user's search purpose, and storing a document ID for specifying each document corresponding to a keyword which is a word included in each document. A document database, a document evaluation database storing an evaluation value that is the relevance of each document with respect to each of the search viewpoints for each search viewpoint and for each document ID, and the search received by the search request receiving unit. A document search unit that searches the document database based on a condition and obtains a document ID that satisfies the search condition; An evaluation value obtaining unit that searches the document evaluation database based on the search viewpoint received by the search request reception unit and obtains an evaluation value of each of the obtained document IDs in the search viewpoint, based on the obtained evaluation value And a search result returning means for determining a document to be output as a search result and returning the result to a user terminal. 2. The user terminal, to which the search result is returned by the search result returning means, sends behavior information including user browsing or evaluation to the search result transmitted from the user terminal as a result of the user browsing the search result. A user action information receiving unit that receives the document ID that is the target of the action and the search viewpoint, a search action database that stores the received document ID, action information, and the search viewpoint; and a search action database that is stored in the search action database. Analyzing a user operation on the document based on the information, calculating an evaluation value of the document in the search viewpoint based on the analysis result, and using the calculated evaluation value for each of the search viewpoint and the document ID in the document evaluation database. 2. A search behavior analysis updating means for updating an evaluation value.
Described information retrieval device. 3. An information retrieval method for retrieving and outputting information conforming to a search request in response to a search request input from a user terminal, the search condition comprising a keyword limiting a search range as the search request. And a search viewpoint for evaluating the priority of information based on the search purpose of the user, and the document database storing a document ID for specifying each document corresponding to a keyword which is a word included in each document is received. A document evaluation in which a search is performed based on a search condition, a document ID that satisfies the search condition is obtained, and an evaluation value that is the relevance of each document with respect to each of the search viewpoints is stored for each search viewpoint and for each document ID. A database is searched based on the obtained document ID and the received search viewpoint, and an evaluation value of each of the obtained document IDs in the search viewpoint is obtained. An information search method comprising: determining a document to be output as a search result based on the obtained evaluation value; and returning the determined document to a user terminal. 4. Behavior information including browsing or evaluation of a user with respect to the search result transmitted from the user terminal as a result of the user browsing the search result at the user terminal to which the search result is returned is targeted for the behavior. Received together with a certain document ID and the search viewpoint, store the received document ID, action information, and search viewpoint in a search action database, and analyze a user's operation on the document based on each information stored in the search action database Calculating an evaluation value of the document in the search viewpoint based on the analysis result, and updating the evaluation value for each of the search viewpoint and the document ID in the document evaluation database with the calculated evaluation value. Item 3. The information retrieval method according to Item 3. 5. An information retrieval program for retrieving and outputting information conforming to a search request in response to a search request input from a user terminal, wherein the search condition includes a keyword for limiting a search range as the search request. And a search viewpoint for evaluating the priority of information based on the search purpose of the user, and the document database storing a document ID for specifying each document corresponding to a keyword which is a word included in each document is received. A document evaluation in which a search is performed based on a search condition, a document ID that satisfies the search condition is obtained, and an evaluation value that is the relevance of each document with respect to each of the search viewpoints is stored for each search viewpoint and for each document ID. A database is searched based on the obtained document ID and the received search viewpoint, and each evaluation value of the obtained document ID in the search viewpoint An information search program characterized in that a document to be output as a search result is determined based on the obtained evaluation value, and returned to the user terminal. 6. Behavior information including browsing or evaluation of a user with respect to the search result transmitted from the user terminal as a result of the user browsing the search result at the user terminal to which the search result is returned. Received together with a certain document ID and the search viewpoint, store the received document ID, action information, and search viewpoint in a search action database, and analyze a user's operation on the document based on each information stored in the search action database Calculating an evaluation value of the document in the search viewpoint based on the analysis result, and updating the evaluation value for each of the search viewpoint and the document ID in the document evaluation database with the calculated evaluation value. Item 5. An information retrieval program according to Item 5. 7. A recording medium storing an information search program for searching for and outputting information matching a search request in response to a search request input from a user terminal, wherein a search range is limited as the search request. A document that receives a search condition that evaluates the priority of information based on search conditions consisting of keywords and a user's search purpose, and stores a document ID that specifies each document in correspondence with a keyword that is a word included in each document A database is searched based on the received search condition, a document ID that satisfies the search condition is obtained, and an evaluation value that is the relevance of each document with respect to each of the search viewpoints is evaluated for each search viewpoint and for each document ID. A document evaluation database stored in the search document based on the acquired document ID and the received search viewpoint. A recording medium storing an information retrieval program characterized by acquiring an evaluation value of each ID, determining a document to be output as a search result based on the acquired evaluation value, and returning the document to a user terminal. 8. Behavior information including browsing or evaluation of a user with respect to the search result transmitted from the user terminal as a result of the user browsing the search result at the user terminal to which the search result is returned is targeted for the behavior. Received together with a certain document ID and the search viewpoint, store the received document ID, action information, and search viewpoint in a search action database, and analyze a user's operation on the document based on each information stored in the search action database Calculating an evaluation value of the document in the search viewpoint based on the analysis result, and updating the evaluation value for each of the search viewpoint and the document ID in the document evaluation database with the calculated evaluation value. Item 7. A recording medium on which the information search program according to Item 7 is recorded. 9. A search operation at a user terminal that evaluates a document by referring to a document obtained as a result of searching a set of a plurality of documents from a document storage unit in response to a search request input from the user terminal at the user terminal. An information retrieval apparatus that obtains an evaluation of the document, calculates an evaluation value for each document based on the obtained evaluation, and uses the evaluation value to order the output of a search result in response to a search request from a user terminal. A document evaluation acquisition unit that acquires the search request, the information specifying the document, and the evaluation of the document from a search action in a user terminal; and a document evaluation storing the search request, the document identification information, and the evaluation of the document. Log storage means, direct evaluation means for calculating a direct evaluation for a document directly evaluated at the user terminal, An indirect evaluation means for calculating an indirect evaluation of a document using a search request from the user terminal and information of the evaluation of the document obtained from the user terminal for the document as a search result for the search request; and the direct evaluation means Document evaluation management means for calculating an overall evaluation value of each document from the direct evaluation and indirect evaluation of each document calculated by the indirect evaluation means, and document evaluation storage means for storing the overall evaluation value corresponding to each document An information retrieval device comprising: 10. The indirect evaluation means uses a search request from a user terminal and information on the evaluation of the document obtained from the user terminal for a document that is a search result for the search request, and directly outputs the information to the user terminal. A partial document acquisition unit that acquires a partial document of interest in a user terminal from a directly evaluated document that is a document that has been automatically evaluated; a document relevance determination unit that determines a document highly relevant to the partial document; Indirect evaluation management means for obtaining an indirect evaluation of the document from the directly evaluated document from the direct evaluation value of the directly evaluated document and the relevance of a certain document to the directly evaluated document. Item 9. The information retrieval device according to Item 9. 11. A search operation in a user terminal which evaluates a document by referring to a document obtained as a result of searching a set of a plurality of documents from a document storage unit in response to a search request input from the user terminal in the user terminal. An information retrieval method for acquiring an evaluation for the document, calculating an evaluation value for each document based on the acquired evaluation, and using the evaluation value to order the output of a search result in response to a search request from a user terminal. Acquiring the search request, the information specifying the document, and the evaluation of the document from a search action in a user terminal, storing the search request, the document specifying information, and the evaluation of the document in a document evaluation log storage unit, The direct evaluation for the document directly evaluated at the user terminal is calculated, and a search request from the user terminal and the search are performed. An indirect evaluation of the document is calculated using information of the evaluation of the document obtained from the user terminal with respect to the document which is a search result for the request, and a total of each document is calculated from the calculated direct evaluation and the indirect evaluation of each document. An information retrieval method, comprising: calculating an evaluation value; and storing the comprehensive evaluation value in a document evaluation storage unit corresponding to each document. 12. The process of calculating the indirect evaluation includes: using a search request from a user terminal and information on the evaluation of the document obtained from the user terminal with respect to a document that is a search result corresponding to the search request; Obtaining a partial document of interest at the user terminal from a directly evaluated document, which is a document directly evaluated at the terminal, determining a document highly relevant to the partial document, and directly evaluating a value of the directly evaluated document 12. The information search method according to claim 11, wherein an indirect evaluation of the document from the directly evaluated document is obtained from the relevance of a certain document to the directly evaluated document. 13. A search operation at a user terminal which evaluates a document by referring to a document obtained as a result of searching a set of a plurality of documents from a document storage means in response to a search request input from the user terminal at the user terminal. An information retrieval program that obtains an evaluation of the document, calculates an evaluation value for each document based on the obtained evaluation, and uses the evaluation value to order the output of search results in response to a search request from a user terminal. Acquiring the search request, the information specifying the document, and the evaluation of the document from a search action in a user terminal, storing the search request, the document specifying information, and the evaluation of the document in a document evaluation log storage unit, The direct evaluation is calculated for the document directly evaluated at the user terminal, and the search request and the search request from the user terminal are calculated. The indirect evaluation of the document is calculated using the information of the evaluation of the document obtained from the user terminal for the document that is the search result for the search request, and each document is calculated from the calculated direct evaluation and indirect evaluation of each document. An information retrieval program characterized by calculating an overall evaluation value of a document and storing the overall evaluation value in a document evaluation storage unit corresponding to each document. 14. The process of calculating the indirect evaluation includes: using a search request from a user terminal and information on the evaluation of the document obtained from the user terminal with respect to a document that is a search result corresponding to the search request; Obtaining a partial document of interest at the user terminal from a directly evaluated document, which is a document directly evaluated at the terminal, determining a document highly relevant to the partial document, and directly evaluating a value of the directly evaluated document 14. The information retrieval program according to claim 13, wherein an indirect evaluation of the document from the directly evaluated document is obtained from the relevance of a certain document to the directly evaluated document. 15. A search operation in a user terminal that evaluates a document by referring to a document obtained as a result of searching a set of a plurality of documents from a document storage unit in response to a search request input from the user terminal and evaluating the document. Recording an information search program for obtaining an evaluation for the document, calculating an evaluation value for each document based on the obtained evaluation, and using the evaluation value to order the output of search results in response to a search request from a user terminal. The retrieval request, the information for identifying the document, and the evaluation of the document from a search behavior in a user terminal, and storing the search request, the document identification information, and the evaluation for the document in a document evaluation log. Means for calculating the direct evaluation of the document directly evaluated at the user terminal, The indirect evaluation of the document is calculated using the information of the evaluation of the document obtained from the user terminal for the search request and the document that is the search result corresponding to the search request, and the direct evaluation of each of the calculated documents is performed. A recording medium recording an information search program, wherein an overall evaluation value of each document is calculated from indirect evaluation, and the overall evaluation value is stored in a document evaluation storage unit corresponding to each document. 16. The processing for calculating the indirect evaluation includes: using a search request from a user terminal and information on the evaluation of the document obtained from the user terminal with respect to a document that is a search result corresponding to the search request; Obtaining a partial document of interest at the user terminal from a directly evaluated document, which is a document directly evaluated at the terminal, determining a document highly relevant to the partial document, and directly evaluating a value of the directly evaluated document 16. The recording medium according to claim 15, wherein an indirect evaluation of the document from the directly evaluated document is obtained from the relevance of a certain document to the directly evaluated document.