JP2002342379A - Method, system and program for ranking popularity of documents and recording medium with the program recorded thereon - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform a popularity ranking of documents, by which even documents having a little information can be highly accurately ranked while being hardly affected by a bias in utilization information. SOLUTION: A document retrieving part 10 stores the utilization information of the users of respective documents on a utilization information database 21. A document ranking system 30 calculates the result score of each of documents from the utilization information to the documents stored on the database, defines the average result score of the partial set of other documents having utilization information including the utilization information on each of documents as an expected score concerning that document, calculates the document score by combination weighting the result score and the expected score with respective significance degrees corresponding to the size of the partial set and stores it on a document score database 22. Document sets receiving a retrieval request from the user are ranked by using the document scores of the respective documents in the document sore database and transmitted to the user.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document search system on a computer network, and more particularly, to a document ranking system which presents a search result list in which the document popularity levels are ranked.

[0002]

2. Description of the Related Art Heretofore, in document search on a computer network, the following methods have been widely used as a method for ranking the popularity of search documents.

(A) A method in which the frequency of occurrence of a word included in a search keyword in a document is used in the descending order of similarity with the search keyword.

(B) A method in which citations from many documents and citations from documents cited in many documents are used in descending order of citations using citation relationships between documents.

However, these methods have the following problems.

(A) Since a search keyword actually input by a user is about several words, sufficient search accuracy cannot be expected.

(B) It is assumed that there is a sufficient amount of citation relation between documents, so that the applicable range is limited.

[0008] As a method free from these problems and restrictions, (a) using the usage information such as the number of references to each document when the user performs a search and the actual evaluation from the user who made the search; A method of calculating an overall “popularity” and performing the calculation based on the calculated “popularity” is used.

[0009]

However, the above prior art has the following problems.

[0010] Generally, the number of pieces of usage information obtained is biased for each document. Also, when a new document is added,
The difference in the number of usage information from other documents is large.

Therefore, if the popularity of a document is determined based only on the usage information for the document, the accuracy will be reduced.

[0012] An object of the present invention is to consider the expected value of the future popularity in addition to the popularity based on the usage information of the conventional document, so that the document is less susceptible to the bias of the information, and the document having less information is used. Another object of the present invention is to provide a method, an apparatus, a program, and a recording medium for ranking popularity of a document, which can rank the documents with high accuracy.

[0013]

According to the present invention, in order to solve the above-mentioned problem, a popularity (document score) is added to a score (actual score) calculated from usage information for a conventional document, and the document is evaluated in the future. An expected value (expected score) that predicts a score that can be obtained in a predictable manner is calculated by the method described below, and ranking is performed based on the calculated document score.

Method of calculating document score Since the distribution of usage information for a document set is considered to represent the characteristics of the document set, usage information according to the characteristics is given to each document in the future. It is considered likely. That is, a highly accurate expected score can be calculated by using the actual score of another document that may have the same use information in the future.

It is considered that the document score has already been determined for a document having a large amount of use information. Therefore, the actual score should be more important than the expected score. It is considered that the expected score should be more important than the actual score.

Based on the above idea, according to the present invention, for a set of documents to be ranked, a performance score of each document is calculated from the usage information for each document by an arbitrary method. A subset of other documents that have usage information that includes the usage information for the document is extracted, and the average performance score of the extracted subset is used as the expected score. A document score is calculated by a weighted combination, and is characterized by the following configuration.

(Invention of Method) (1) A method of ranking the popularity of documents that presents a search result list in which the degrees of popularity of documents are ranked when searching for documents on a computer network. Store usage information in the usage information database,
For a set of documents to be ranked and stored in the database, a performance score of each document is calculated by an arbitrary method from usage information for each document, and each document includes usage information for the document. The average result score of a subset of other documents having such usage information is set as an expected score, and the document score is calculated by a combination of the size of the subset and the result score and the expected score weighted by their importance. The method is characterized in that the document set is stored in a document score database, the document set received a search request from the user is ranked using the document score of each document in the document score database, and transmitted to the user. .

(2) In the calculation of the document score, the actual score of each document is calculated by an arbitrary method using the usage information extracted from the usage information database, and each document is calculated using the usage information and the actual score. Is calculated, and the document score of each document is calculated from the performance score, the expected score, and the importance.

(3) The calculation of the expected score and the importance is performed by selecting an arbitrary document for which the expected score has not been calculated, and selecting another document having usage information that the selected document can have in the future. A set of documents is derived, an average performance score of the documents included in the set is calculated as an expected score, and importance of the expected score is calculated from the size of the set.

(System Invention) (4) A document popularity ranking system which presents a search result list in which document popularity is ranked when a document is searched on a computer network. A usage information database that stores usage information, and for a set of documents to be ranked stored in the database, an actual score of each document is calculated by an arbitrary method from usage information for each document, For each document, the expected score is the average performance score of a subset of other documents that have usage information that includes usage information for that document. A document ranking system that calculates a document score with a combination weighted by degrees and stores it in the document score database A document search unit that ranks a set of documents requested to be searched by a user using the document score of each document in the document score database and transmits the document set to the user. I do.

(5) The document ranking system calculates a performance score of each document by an arbitrary method using the usage information extracted from the usage information database, and returns a performance score calculation unit to the document score calculation unit. An expected score calculation unit that calculates an expected score and importance of each document using the use information and the performance score and returns the calculated score and importance to the document score calculation unit; A document score calculation unit that calculates a document score and stores the calculated document score in the document score database.

(6) The document ranking system includes a means for selecting an arbitrary document for which the expected score has not been calculated, and a method for selecting another document having use information that the selected document can have in the future. Means for deriving a set, means for calculating an average performance score of documents included in the set to obtain an expected score, and means for calculating importance of the expected score from the size of the set. And

(Invention of Program) (7) The program is characterized in that it has a processing procedure for causing a computer to execute the document popularity ranking method.

(Invention of Recording Medium) (8) A program for causing a computer to execute the processing procedure in the above-mentioned document popularity ranking method is recorded on a computer-readable recording medium. .

[0025]

FIG. 1 is a system configuration diagram showing an embodiment of the present invention. In this figure, the system is:
Document search unit 10 and usage information DB (database) 21
, Document score DB 22, and document ranking system 3
Has 0.

The document ranking system 30 has a document score calculation unit 31, a performance score calculation unit 32, and an expected score calculation unit 33.

The document search unit 10 receives a search request from a user, extracts a set of documents that meet the request by any method, and calculates a document score of each document in the document score DB 22 for the set of documents. Use it to rank and send to users. Further, after the search, usage information such as reference to the document and evaluation from the user is acquired, and the usage information DB 21
To accumulate.

The usage information DB 21 is a database storing usage information such as the number of presentations of each document to the user, the number of references from the user, the number of evaluations, and evaluation points. The document score DB 22 is a database that stores a document score for each document.

In the document ranking system 30, the document score calculation unit 31 extracts usage information such as the number of presentations, the number of references, the number of evaluations, and the evaluation points of each document stored in the usage information DB 21. Then, the actual score and the expected score of each document are obtained by using the expected score calculation unit 33, and the document score of each document is calculated from the obtained score and stored in the document score DB 22.

The performance score calculation unit 32 is called from the document score calculation unit 31 and displays the number of presentations, the number of references,
The result score of each document is calculated from the usage information such as the number of evaluations and the evaluation points, and is returned to the document score calculation unit 31.

The expected score calculation unit 33 is called from the document score calculation unit 31 and calculates the number of presentations, the number of references,
The expected score of each document and its importance are calculated from the usage information such as the number of evaluations and evaluation points and the performance score, and are returned to the document score calculation unit 31.

FIG. 2 is a flowchart showing the operation of the document ranking system 30. Referring to FIG.
In 1, the document score calculation unit 31 extracts usage information such as the number of presentations, the number of references, the number of evaluations, and the evaluation points of each document from the usage information DB 21.

In step S42, the document score calculation unit 31
Handed over the extracted usage information to the performance score calculation unit 32, and the performance score calculation unit 32 calculates the performance score of each document by using an arbitrary method using the obtained usage information.
Return to 1.

In step S43, the document score calculation section 32
Handovers the usage information and the performance score to the expected score calculation unit, and the expected score calculation unit 33 calculates the expected score and the importance of each document by using them, and returns them to the document score calculation unit 31. An example of a method of calculating the expected score and the importance will be described with reference to FIG.

In step S44, the document score calculation unit 31
Calculates the document score of each document from the actual score, the expected score, and the importance. For example, when the importance takes a value of 0 to 1, (1—one importance) × actual score + importance × expected score.

In step S45, the document score calculation unit 31
Stores the calculated document score of each document in the document score DB 22
To store.

FIG. 3 shows a flowchart of the calculation of the expected score and its importance performed by the calculation unit 33. In the figure, a step S51 selects an arbitrary document for which an expected score has not been calculated.

In step S52, a set of other documents having use information that the selected document can have in the future is derived. For example, a set of documents satisfying the following conditions is derived.

The number of presentations is large. The number of references is large. The number of evaluations is large. “Difference in number of presentations> Difference in number of references> Difference in number of evaluations”. Calculate the average performance score and use it as the expected score.

In step S54, the importance of the expected score is calculated from the size of the set obtained in step S52. For example, the number of documents in the document set obtained in step S52 ÷ the number of all documents.

Step S55 is a branch. If the expected scores have been calculated for all the documents, the process ends. Otherwise, the process returns to step S51.

According to the present invention, a part or all of the processing functions of the apparatus shown in FIG. 1 are implemented as a program and realized by using a computer, or FIG. 2 or FIG.
Can be configured as a program and executed by a computer. In addition, a program for realizing the processing function of each unit in the computer or a program for causing the computer to execute the processing procedure is stored in a recording medium readable by the computer, for example, FD (Floppy Disk: registered trademark),
It can be recorded on an MO, ROM, memory card, CD, DVD, removable disk, etc., stored or provided, and distributed via a communication network such as the Internet. It is possible.

[0043]

As described above, according to the present invention, the degree of popularity (document score) is added to the score (actual score) calculated from the usage information for the conventional document, and the score that the document can take in the future is obtained. Calculates the expected value (expected score) that predicts and ranks the documents based on the calculated document scores, so that the document is hardly affected by the bias of information, and a high-precision ranking is performed even for a document with little information. It has the effect of being able to do so.

[Brief description of the drawings]

FIG. 1 is a diagram showing a system configuration according to an embodiment of the present invention.

FIG. 2 is a flowchart of the embodiment.

FIG. 3 is a flowchart of calculating an expected score and its importance in the embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Document search part 21 ... Usage information database 22 ... Document score database 30 ... Document ranking system 31 ... Document score calculation part 32 ... Actual score calculation part 33 ... Expected score calculation part

Claims (8)

[Claims] 1. A document popularity ranking method for presenting a search result list in which document popularity is ranked at the time of document search on a computer network, wherein a usage information of a user of each document is used in a usage information database. For a set of documents to be ranked which is stored in the database, an actual result score of each document is calculated from the usage information for each document by an arbitrary method. The average performance score of a subset of other documents having usage information that includes usage information for the expected score is defined as an expected score, and the performance score and the expected score are weighted according to their importance according to the size of the subset. A document score is calculated and stored in a document score database. Popularity ranking of documents, characterized by transmitting using a document score of each document that is in the score database performs ranking to a user. 2. The document score is calculated by using the usage information extracted from the usage information database to calculate a performance score of each document by an arbitrary method, and using the usage information and the performance score to calculate the document score of each document. The method according to claim 1, wherein an expected score and an importance are calculated, and a document score of each document is calculated from the performance score, the expected score, and the importance. 3. The calculation of the expected score and the degree of importance is performed by selecting an arbitrary document for which the expected score has not been calculated, and selecting another document having usage information that the selected document can have in the future. The average score of documents included in the set is calculated as an expected score, and the importance of the expected score is calculated from the size of the set. How documents are ranked by popularity. 4. A document popularity ranking system that presents a search result list in which document popularity is ranked when a document is searched on a computer network, wherein the system stores user usage information of each document. A usage information database to be stored, and for a set of documents to be ranked stored in the database, a performance score of each document is calculated by an arbitrary method from usage information for each document, and for each document, The expected score is the average performance score of a subset of other documents having usage information that includes usage information for, and the combination of the performance score and the expected score weighted by their importance according to the size of the subset. A document ranking system that calculates the document score and stores it in the document score database, and receives search requests from users A document search unit that ranks the received document set by using a document score of each document in the document score database and transmits the document set to a user. system. 5. The document ranking system, wherein a performance score of each document is calculated by an arbitrary method using usage information extracted from the usage information database, and a performance score calculation unit is returned to the document score calculation unit. An expected score calculation unit that calculates an expected score and importance of each document using the usage information and the performance score and returns the calculated score and importance to the document score calculation unit; and a document of each document based on the performance score, the expected score, and the importance. The document popularity ranking system according to claim 4, further comprising: a document score calculation unit that calculates a score and stores the calculated document score in the document score database. 6. The document ranking system, comprising: means for selecting an arbitrary document for which the expected score has not been calculated; and a set of other documents having use information that the selected document can have in the future. Means for calculating the average performance score of the documents included in the set to obtain an expected score, and means for calculating the importance of the expected score from the size of the set. The document ranking system according to claim 4 or 5, wherein 7. A document popularity ranking method, comprising a processing procedure for causing a computer to execute the document popularity ranking method according to any one of claims 1 to 3. program. 8. A program for causing a computer to execute the processing procedure in the document popularity ranking method according to claim 1, wherein the program is recorded on a computer-readable recording medium. A recording medium that records a method of ranking popularity of documents characterized by the following.