JP2003186912A - ADAPTATION CALCULATION METHOD AND DEVICE FOR Web PAGE RETRIEVAL, RETRIEVAL RESULT DISPLAY METHOD AND DEVICE FOR Web PAGE RETRIEVAL, AND PERFORMANCE PROGRAM FOR THE SAME METHOD AND RECORDING MEDIUM WITH ITS PROGRAM RECORDED - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To perform the keyword retrieval of a Web page retrieval system by outputting retrieval results by page set units in the highly precise order adapted to a keyword. <P>SOLUTION: In a client terminal 3, a retrieving means 31 performs keyword retrieval to a Web page retrieval system 1, and receives the N pieces of scores and URL of hit pages from the retrieval system 1, and a page DB access means 32 obtains the depth of the page set ID and page set tree structure of each page from a route from a page DB 2, and registers the page set ID in a hit page set list, and a score calculating means 33 calculates a score indicating the adaptation to the keyword based on the depth of the page set tree structure and the score. A hit page set list generating means 34 uses the score as the score element of the page set ID of the list, or adds it to an already registered score element, and a sort means 35 sorts the generated hit page set list with the score element, and outputs it. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and apparatus for calculating the degree of matching (fitness) with a keyword when inquiring a Web page search system with a keyword and searching page information. ,
The present invention relates to a method and apparatus for displaying search results with high accuracy in page set units in descending order of matching with keywords when a large number of search results by keywords are obtained as page information.

[0002]

2. Description of the Related Art In the WWW (World Wide Web), a client terminal can make an inquiry using a keyword to a Web page search system, and information about a page hit by the keyword can be searched. Conventionally, when a plurality of search results are obtained by such a Web page search system, the list display of the search results is generally output without being sorted, sorted, and processed by the search system. But,
The page information obtained by such a web page search is usually enormous, and in order to know which is the search result that truly matches the query by the keyword, the URL ( It took a lot of time and effort to go through each Universal Resource Locator) and find out by referring to the page set of the site.

In order to improve such a situation, attempts have been made to rank and present search results. For example, in the technique disclosed in Japanese Patent Laid-Open No. 2001-60165, pages existing on the server are collected in advance and the link reference status across the servers is acquired.
The link reference situation is used to determine the importance of each page, and the search results are sorted and output according to the importance.

[0004]

[Patent Document 1] Japanese Unexamined Patent Application Publication No.
In the technology disclosed in 001-60165, even if the importance level of a page is determined from the link reference status across each server, it is irrelevant to the fact that it is suitable for the keyword in the Web page search by the keyword. There was a big problem that the search results were not sorted in the order of matching the keywords.

Therefore, after all, a great deal of time and effort is required to search for a desired page by referring to the URLs shown in a large number of search results one by one and referring to the page set of the site. There was no change.

The present invention provides a matching degree calculating method and apparatus capable of highly accurately calculating the degree of matching with a keyword from a search result with respect to a query by a keyword for a Web page search system. It is an object of the present invention to provide a fitness degree display method and device capable of displaying a list of search results in a highly accurate ranking that matches a keyword.

[0007]

In order to solve the above problems, the present invention provides a process of obtaining a score and a URL of a page hit by a keyword inquiry to a Web page search system, and accessing a page database. Obtaining a page set ID to which the page belongs and depth information from the root of the page set tree structure, and calculating a score representing a goodness of fit to the inquiry based on the depth information of the page set tree structure and the score. And a step of performing the following.

Alternatively, the process of obtaining the score and URL of the page hit by the keyword inquiry to the Web page search system, the page set ID to which the page belongs and the depth information from the root of the page set tree structure, and the page. A method of calculating a fitness level for Web page retrieval, comprising: a step of calculating a score representing a fitness level for the inquiry based on depth information of a set tree structure and the score.

Alternatively, in the process of calculating the score, there is a function that the score becomes higher as the page score becomes higher, and the score is calculated from a calculation formula in which the page closer to the root of the page set tree structure has the higher score. Web characterized by
The method of calculating the fitness of page search is used.

Alternatively, a process of obtaining N scores and URLs of pages hit by an inquiry to a Web page search system by a keyword, and accessing a page database to access the page set ID and page set to which each of the N pages belongs. A step of obtaining depth information from the root in the tree structure, a step of calculating a score representing the suitability for the inquiry based on the depth information and the score of the page set tree structure for each page, and for each page The page set ID in the hit page set list
Is registered and the calculated score is set as the score element of the page set ID of the hit page set list, or if the page set ID is already registered, the calculated score is added to the score element. And a step of sorting the hit page set list by the value of the score element and outputting the list.

Alternatively, the process of acquiring the score and URL of the page hit by the inquiry by the keyword to the Web page search system, the page set ID to which the page belongs and the depth information from the root in the page set tree structure, and each page. In the process of calculating a score representing the fitness to the inquiry based on the depth information of the page set tree structure and the score, and registering the page set ID in the hit page set list for each page, and the calculated score Is set as the score element of the page set ID in the hit page set list, or if the page set ID is already registered, the step of adding the calculated score to the score element, and the hit page set list And the process of sorting by the value of the score element and outputting, To, and means the search result display method of a Web page search, characterized in that.

Alternatively, in the process of calculating the score, the score is calculated from a formula having a function in which the higher the score of the page, the higher the score, and the closer to the root of the page set tree structure, the higher the score. W characterized by
The search result display method of the eb page search is used as a means.

Alternatively, a means for acquiring the score and URL of the page hit by the keyword inquiry to the Web page search system, and accessing the page database, the page set ID to which the page belongs
And means for obtaining depth information from the root in the page set tree structure, and means for calculating a score representing the suitability for the query based on the depth information and the score of the page set tree structure, W characterized by
The fitness calculation device for the eb page search is used as a means.

Alternatively, the score and URL of the page hit by the keyword inquiry to the Web page search system, the page set ID to which the page belongs and the depth information from the root in the page set tree structure, and the page. A fitness calculation device for Web page search, comprising: depth information of a set tree structure; and means for calculating a score representing the fitness for the inquiry based on the score.

Alternatively, a means for acquiring N scores and URLs of pages hit by a keyword inquiry to the Web page search system and a page database to access the page set ID and page set to which each of the N pages belongs. Means for obtaining depth information from the root in the tree structure, means for calculating a score representing the suitability for the inquiry based on the depth information and the score of the page set tree structure for each page, and for each page The page set ID in the hit page set list
Means for setting the calculated score as a score element of the page set ID of the hit page set list, or adding the calculated score to the score element if the page set ID is already registered When,
Means for sorting the hit page set list by the value of the score element and outputting the sorted result.
The search result display device on page b is used as a means.

Alternatively, a score and URL of a page hit by a keyword inquiry to the Web page search system, a means for acquiring the page set ID to which the page belongs and depth information from the root in the page set tree structure, and each page. A means for calculating a score representing a goodness of fit for the inquiry based on the depth information of the page set tree structure and the score; and registering the page set ID in a hit page set list for each page, and calculating the score Is set as the score element of the page set ID of the hit page set list, or when the page set ID is already registered, means for adding the calculated score to the score element, and the hit page set list And the means to output by sorting by the value of the score element, To, and means the search results display of a Web page search, characterized in that.

Alternatively, a program for executing a fitness calculation method for Web page search for causing a computer to execute the steps in the fitness calculation method for Web page search described above is used as means.

Alternatively, a program for executing a search result display method for Web page search for causing a computer to execute the above-described process in the search result display method for Web page search is used as means.

Alternatively, a program for causing a computer to execute the steps in the method for calculating the fitness for Web page search described above is recorded in a computer-readable recording medium, and the fitness for Web page search is characterized. A recording medium recording an execution program of the calculation method is used as a means.

Alternatively, a search result display method for a web page, characterized in that a program for causing a computer to execute the steps in the search result display method for the web page search is recorded in a computer-readable recording medium. A recording medium recording an execution program of the display method is used as a means.

According to the present invention, when a page search result is obtained in the Web page search system, the score of the page reflecting the weight of the keyword, the page set ID to which the page belongs and the root (root) in the tree structure of the page set. ), The score as the goodness of fit for the keyword is calculated with high accuracy by calculating the score reflecting the score and the depth information. When multiple page search results are obtained, the page set ID is sorted by the score element obtained by adding the score for each page for each page set ID, and the page search units are sorted in order of matching the search results with the keyword. To output.

[0022]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described in detail below with reference to the drawings. First, terms used in the present invention will be described.

Ranking is a process of sorting and outputting page sets in order of suitability.

A page set is the root of a tree structure.
It means a page set that has a top page that has the following path, and the path from the root to each page belonging to the page set is uniquely determined (each page corresponds to a node of a tree structure). A site is a form of this page set.

The depth in the tree structure of the page set is
A page shows the depth from the root of the tree structure of a page set. For example, if a page is reachable by following the link twice from the root in the page set tree structure, the depth of this page in the page set tree structure is 2
That is. The depth of the route is 0.

The Web page search system is an indexer that extracts a keyword from a set of Web pages to be searched to create an index, and a search request is scored in consideration of the weight of a word (keyword). , Pages with search results in descending order of score (This is a system consisting of a search server that can return a specified number of pages that hit the search ("Information Retrieval and Language Processing", Takenobu Tokunaga, The University of Tokyo Press) ISBN4-13-
065405-5 (hereinafter referred to as document [1]), pp,
11-65).

The score is a numerical value of the degree of conformity with respect to a page output by a certain web page search system. The score calculation method should be given mainly by a function based on the product of the number of keywords included in the page (tf) and the reciprocal of the number of pages that include the keyword in the search target page set (idf). Many (Reference [1] p
p. 26-32).

Next, FIG. 1 shows an example of a system configuration for explaining the first embodiment of the present invention. In FIG.
Reference numeral 1 is a web page search system, 2 is a page DB (database), and 3 is a client terminal. The connection between these components is arbitrary as long as necessary data can be exchanged between the respective parts.

The Web page search system 1 is provided with an index DB 11 that stores indexes in which keywords have been extracted from the collected pages and a search server having a search engine. The DB 11 is searched and the information of the page set hit by the keyword is transmitted to the client terminal 3. Page DB2 is all pages I
D, the page set ID to which the page belongs, and the depth information from the tree structure of this page set are accumulated.

The client terminal 3 makes an inquiry to the Web page search system 1 using a keyword,
The search means 31 for receiving N scores and URLs of pages hit by the keyword inquired from the b page search system 1 and the page DB 2 are accessed to access the page set ID and the root of the page set tree structure to which each N page belongs. Page DB access means 32 for registering the page set ID obtained in the hit page set list, and the degree of conformity to the keyword based on the depth information and score of the obtained page set tree structure. And a score calculation means 33 for calculating a score representing the calculated score, and the calculated score as page set I in the hit page set list.
The hit page set list generation means 34 is set as the score element of D, or when the page set ID is already registered, the calculated score is added to the score element.
And a sort means 35 for sorting the generated hit page set list by the value of the score element and outputting the sorted result.

FIG. 2 shows a flowchart of the ranking process according to the embodiment of the present invention realized by the above-mentioned configuration. S1 to S8 are processing steps.

First, a keyword is input to the Web page search system 1 from the client terminal 3 to perform an inquiry search, and information about a page hit by the keyword from the Web page search system 1 is obtained, along with the score and URL of the page. N items are received (S1). Here, the score is W as described above.
This is calculated on the eb page search system 1 side in consideration of the weight of the word (keyword).

Next, the page set DB2 is accessed from the client terminal 3 by the obtained URL, the page set ID of each N pages and the depth information in the page set tree structure are acquired (S2), and the hit page set is obtained. Empty the list (S3).

Next, in loop 1, the following processing is repeated for each page for all N hit pages.

First, based on the acquired score and the depth information of the page set tree structure, the score of the page as the matching degree to the search keyword is calculated (S4). Where the page is p
When the score of this page is score (p), the score of each page hit is s (p), and the depth value in the tree structure of the page set is d (p). expressed.

Score (p) = f (s (p)) × (1 / (α
+ (D (p)) ² × β))) where α and β represent constants larger than 0 and smaller than 1, and f
Represents an arbitrary function such that when s (p) is high, score (p) is also high.

Next, when the page set ID of this page is not registered in the hit page set list (S5),
The page set ID of this page is registered in the hit page set list, and the calculated score is set as the score element of the page set ID (S6).

Here, when the page set ID is already registered during the repeated processing of N pages by the loop 1 (S5), the calculated score is added to the score element (S7).

Finally, the hit page set list is sorted by point elements and output (S8), and the process ends.

FIG. 3 shows an example of high / low scores calculated according to the present invention in the case where the page set takes the form of a site. In this case, the hit page set list obtained by the present invention becomes a hit site list, and the search result can be obtained for each site.

In FIG. 3A, in the site structure having a certain page set ID, the page hit by the search is
Since the page is the root top page (depth value 0) and the first hierarchy page (depth value 1) close to it, the score element of the site has a high value. On the other hand, in FIG. 3B, in the site structure having a certain page set ID, the page hit by the search is the page of the second hierarchy (depth value 2) deep from the top page that is the root, so that site The scoring element of is a low value. The score (p) above
As is clear from the calculation formula of, the score element has a higher value for a site including many pages with a higher page search score.

As a second embodiment of the present invention, a configuration in which the page set ID for each page stored in the page DB and the depth information in the page set tree structure are included in the index DB is considered. To be in this case,
The page DB access means of the client terminal 3 becomes unnecessary, and the search means 31 acquires these pieces of information and registers the page set ID in the hit page set list. Other configurations and ranking processing are the same as those in the first embodiment.

Further, as a third embodiment of the present invention, it is conceivable that the Web page search system side has means for performing ranking processing. In this case, the page DB access means (not required in the second embodiment), the score calculation means, the hit page set list generation means, the sorting means, etc. in the first and second embodiments are on the Web page search system side. Will have. The ranking process realized by these means is the same as in the first and second embodiments. The client terminal has only the search means, and receives the hit page set list from the sort means as the search result.

The means for realizing the functions shown in FIG. 1 or the like may be configured by a computer program, or the processing procedure shown in FIG. 2 or the like may be configured by a computer program, and the program is stored in the computer. Needless to say, the program can be executed by a computer, or a program for causing a computer to execute the processing procedure can be read by a computer-readable recording medium such as an FD (floppy disk). (Registered trademark)) and M
The data can be recorded in O, ROM, memory card, CD, DVD, removable disk, etc., and can be saved or distributed. It is also possible to provide the above program through a network such as the Internet or electronic mail. The present invention can be implemented by installing the above program in the computer from these recording media, or by downloading from the network and installing the above program in the computer.

[0045]

As described above, according to the present invention, in the Web page search system, when a page search result is obtained, the score of the page, the page set ID and the page set tree to which the page belongs. By acquiring the depth information from the root in the structure and calculating the score reflecting the score and the depth information, the score as the goodness of fit for the keyword can be calculated with high accuracy. Further, when a plurality of search results are obtained, the page set IDs are sorted by the score element obtained by adding the score for each page for each page set ID, and the search result is matched with the keyword in a highly accurate order. Can be output in page set units.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a system for realizing an exemplary embodiment of the present invention.

FIG. 2 is a flowchart showing a ranking process according to an exemplary embodiment of the present invention.

3 (a) and 3 (b) are diagrams for explaining high / low of a score as a goodness of fit in the present invention.

[Explanation of symbols]

1 ... Web page search system 11 ... Index DB 2 ... Page DB 3 ... Client terminal 31 ... Search means 32 ... Page DB access means 33 ... Score calculation means 34. Hit page set list generation means 35 ... Sorting means

Claims (14)

[Claims] 1. A process of obtaining a score and URL of a page hit by a keyword inquiry to a Web page search system, and accessing a page database to obtain a page set ID to which the page belongs and a root in a page set tree structure. A step of obtaining depth information, and a step of calculating a score representing a goodness of fit to the query based on the depth information of the page set tree structure and the score, Fitness calculation method. 2. A process of obtaining a score and URL of a page hit by a keyword inquiry to the Web page search system, a page set ID to which the page belongs and depth information from the root of the page set tree structure, and the page. A method of calculating a fitness level for Web page search, comprising: a step of calculating a score representing a fitness level for the inquiry based on the depth information of the set tree structure and the score. 3. The process of calculating the score has a function in which the score increases as the page score increases,
The score is calculated from a calculation formula in which the score is higher as the page is closer to the root of the page set tree structure. The fitness calculation method for Web page search according to claim 1, wherein the score is calculated. 4. A process of acquiring N scores and URLs of pages hit by a keyword inquiry to a Web page search system, and accessing a page database to access a page database and a page set ID and a page set to which each N page belongs. A step of obtaining depth information from the root in the tree structure, a step of calculating a score representing a goodness of fit to the query based on the depth information of the page set tree structure and the score for each page, and The page set ID is registered in the hit page set list, and the calculated score is set as the score element of the page set ID in the hit page set list, or the page set ID is already calculated when the page set ID is already registered. The process of adding the score obtained to the score element, A step of outputting the sorted values of the scoring element list has, the search result display method of a Web page search, characterized in that. 5. A process of obtaining a score and URL of a page hit by a keyword inquiry to the Web page search system, a page set ID to which the page belongs and depth information from the root of the page set tree structure, and each page. In the process of calculating the score indicating the goodness of fit to the query based on the depth information of the page set tree structure and the score, the page set ID is registered in the hit page set list for each page, and the calculated score is calculated. As a score element of the page set ID of the hit page set list, or adding the calculated score to the score element when the page set ID is already registered, and the hit page set list And a process of sorting by the value of the score element and outputting, A method for displaying a search result of a web page search, which is characterized in that: 6. The process of calculating a score has a function in which the score increases as the page score increases,
6. The search result display method of the Web page search according to claim 4, wherein the score is calculated from a calculation formula in which the score is higher as the page is closer to the root of the page set tree structure. 7. A means for obtaining the score and URL of a page hit by a query by a keyword to a Web page search system, and accessing a page database to obtain a page set ID to which the page belongs and a root in a page set tree structure. A means for acquiring depth information; and means for calculating a score representing the suitability for the inquiry based on the depth information of the page set tree structure and the score. Fitness calculator. 8. A means for acquiring a score and URL of a page hit by a keyword inquiry to a Web page search system, a page set ID to which the page belongs and depth information from a root in a page set tree structure, and the page. A fitness calculation device for Web page search, comprising: depth information of a set tree structure; and means for calculating a score representing the fitness for the inquiry based on the score. 9. A means for obtaining N scores and URLs of pages hit by a keyword inquiry to a Web page search system, and a page database accessed to access a page database and a page set ID and a page set to which each N page belongs. Means for obtaining depth information from the root in the tree structure, means for calculating a score representing the suitability for the inquiry based on the depth information of the page set tree structure and the score for each page, and for each page The page set ID is registered in the hit page set list, and the calculated score is set as the score element of the page set ID in the hit page set list, or the page set ID is already calculated when the page set ID is already registered. Means for adding the score to the score element, and the hit page set list And means for outputting the door sorted by the value of the score elements have, search result display of the Web page, characterized in that. 10. A means for acquiring a score and URL of a page hit by a keyword inquiry to the Web page search system, a page set ID to which the page belongs and depth information from the root of the page set tree structure, and each page. A means for calculating a score representing a goodness of fit to the inquiry based on the depth information of the page set tree structure and the score; and registering the page set ID in a hit page set list for each page, and calculating the score Is set as the score element of the page set ID of the hit page set list, or if the page set ID is already registered, means for adding the calculated score to the score element, and the hit page set list And means for outputting the value by sorting by the value of the score element, A search result display device for Web page search, characterized in that. 11. A process in a method for calculating a fitness level for Web page search according to claim 1,
An execution program of a fitness calculation method for Web page search, which is a program to be executed by a computer. 12. A web page search search, characterized in that a program for causing a computer to execute the steps in the search result display method of a web page search according to any one of claims 4 to 6. Execution program of result display method. 13. A process in a method for calculating a fitness level for Web page search according to claim 1,
A recording medium having a program to be executed by a computer, the program being recorded on a computer-readable recording medium, the recording medium storing an execution program of a method for calculating a fitness level for Web page search. 14. A program for causing a computer to execute the steps in the search result display method of a web page search according to claim 4, wherein the program is readable by the computer. A recording medium storing an execution program of a search result display method for a web page search, which is recorded in the above.