JP2014075064A - Web search device, web search method and program - Google Patents

Abstract

The order of presenting search results to a user is close to the target result of the user.
A document-to-word evaluation score part that collects Web pages and initializes document-to-word evaluation scores on the Web page, a search request receiving part that receives a specified search keyword, and a specified search keyword Refer to the keyword importance level determination unit that determines the importance level for each search keyword and the storage unit that stores the action vs. evaluation-based score indicating the correspondence between the operation content and the added point added to the importance level for each search keyword. And a document-to-word evaluation score update unit that updates the action-to-evaluation-based score for each Web page based on the operation content and the added points of the search keyword.
[Selection] Figure 1

Description

  The present invention relates to improving the quality of search results in a Web (hereinafter referred to as Web) search.

  A system for searching for a target document from a large number of documents on the Web every day is indispensable.

  In the early days of the Web, there were directory-type search systems where operators manually organized Web pages. However, the amount of Web documents and the pace of increase have increased, making it impossible to keep up. The mainstream is a system that collects the above documents using a program that automatically collects, analyzes the collected documents, automatically creates an index, and presents search results to the user based on the index.

  At this time, in order for the user of the search system to efficiently reach the target document, the pages are ranked using a specific algorithm and presented to the system user in descending order of the rank for convenience. The system is common.

  There is the following Patent Document 1 as a known technical document related to the present invention.

  Patent document 1 is a method in a computing system that ranks Web pages in a result set, and enumerates as a result in a result set generated in response to a query that includes one or more query term sets. The query terms reflecting the level of relevance of the web page to the web page (a) and the query term (b) in response to content-related user interaction with the web page being As a result in a set of individual results generated in response to an individual query including a step of adjusting an evaluation score for each combination of the query terms in the set of For each of the listed web pages, one of the web page (a) and the individual set of query terms (b) Selecting an evaluation score adjusted for each combination with a query term; and determining a ranking score for the web page in the individual query result by combining the selected evaluation score. A method characterized by this is disclosed.

Japanese Patent Application No. 2009-503249

  The means for determining the page rank in Patent Document 1 is approximately the following operation. A “page that finds a page that matches the purpose” is performed by the user for a result corresponding to a web page included in a set of search result lists generated in response to a search request including one or a plurality of search keyword sets. In response to each of a number of possible content interactions (for example, printing the content, registering in favorites, etc.), the system selects one keyword from the web page and search keyword pair. The evaluation score is adjusted (increased) for each combination.

  As a result of repeating such user interaction for each Web page, for each of the Web pages included in the individual result set generated in response to the individual search request including one set of individual search keywords, The evaluation score of keywords that are strongly related to the Web page will increase. In addition, the ranking of the Web page in response to the user's search request is determined in descending order of the total evaluation score for the search keyword calculated for each Web page in the set of search keywords. Therefore, by repeating the evaluation of the user's search result to the Web page, the Web page more strongly related to the search keyword is presented to the user as a high-rank page, so that the quality of the search result list is improved.

  By the way, the following problems may occur in this system.

  For example, suppose you have a very major programming language A and a minor programming language B. In addition, Web page A contains the language specification document for programming language A and the words "programming language A", "programming language B", and "specification". Web page B contains the language specification document for programming language B. And “programming language A”, “programming language B”, and “specification” as words. In this case, the following values are assumed to simplify the initial value of the evaluation score for each Web page.

(1) Web page A: Evaluation score for keyword “programming language A” = 0, evaluation score for keyword “programming language B” = 0, evaluation score for keyword “specification” = 0
(2) Web page B: Evaluation score for keyword “programming language A” = 0, evaluation score for keyword “programming language B” = 0, evaluation score for keyword “specification” = 0

  Here, for the purpose of investigating the specification of programming language A or B, the user must specify “programming language A” and “specification” as a search keyword or “programming language B” and “specification” as a pair. Think. Since programming language A is larger than programming language B, a user search request for programming language A (search by keyword `` programming language A '' or `` specification '') is generally a search request for programming language B (keyword `` programming language B ”and“ specification ”) are expected to be more. The increase in the number of search requests by the user naturally increases the chance that the Web page is selected and the evaluation score increases.

  Here, (number of times Web page A is evaluated with the search keywords “programming language A” and “specification”): (number of times web page B is evaluated with the search keywords “programming language B” and “specification”) = 10 If it is: 1, the final Web page evaluation points will be as follows.

(1) Web page A ...
Evaluation score for keyword “programming language A” = 10a,
Evaluation score for keyword “programming language B” = 0a (not evaluated due to low relevance), evaluation score for keyword “specification” = 10a
(2) Web page B ...
Evaluation score for keyword “programming language A” = 0a (not evaluated due to low relevance), evaluation score for keyword “programming language B” = 1a,
Evaluation score for keyword "specification" = 1a

  Here, for the purpose of investigating the specification of programming language B, it is assumed that a search is performed by specifying “programming language B” and “specification” as search keywords.

  At this time, the sum of the evaluation scores for determining the ranking of Web page A and Web page B is as follows.

(1) Web page A evaluation score total = “programming language B” evaluation score 0a + “specification” evaluation score 10a = 10a
(2) Web page B evaluation score total = "programming language B" evaluation score 1a + "specification" evaluation score 1a = 2a

  As a result, "Web Page A Evaluation Score Total"> "Web Page B Evaluation Score Total", and Web page A is presented with priority even though the user wants to know the programming language B specifications. Therefore, there arises a problem that the quality of the search result is deteriorated.

  An object of the present invention is to provide a system that improves the accuracy of determining a page rank and presents a search result that is easier to use for a user.

  According to one aspect of the present invention, a document-to-word evaluation score unit that collects a Web page and initializes a document-to-word evaluation score of the Web page, a search request receiving unit that receives a specified search keyword, Stores an action vs. evaluation base score indicating a correspondence relationship between a keyword importance determining unit that determines the importance for each search keyword from the specified search keyword and an operation content and a point added to the importance for each search keyword. A document-to-word evaluation score updating unit that updates the action-to-evaluation base score for each Web page based on the operation content and the added points of the search keyword with reference to a storage unit; A web search device is provided.

  Importance score for each keyword determined by the above-mentioned means when adding score for each keyword to be added to keyword evaluation score to update document vs. word evaluation score of web page from user content correspondence (evaluation) As a weight, the evaluation score of a keyword that characterizes a Web page (a keyword that represents an information category) is likely to increase. A general word that exists in any category is less likely to be evaluated. In addition, even for a minor word that is not often searched, an important keyword for the document is easily evaluated.

  The document-to-word evaluation score update unit updates the action-based evaluation base score for each Web page based on a subtraction point indicating a search keyword for subtracting the weight of the search keyword for each Web page. Features.

  The document-to-word evaluation score update unit updates the document-to-word evaluation score for each Web page at a certain time interval, and if there is a deleted Web page document, the document-to-word evaluation score The process of deleting from is continued.

  According to another aspect of the present invention, a document-to-word evaluation scoring step for collecting a web page and initializing a document-to-word evaluation score for the web page; a search request receiving step for receiving a designated search keyword; , Memorize action vs. evaluation-based score indicating the correspondence between the keyword importance determination step for determining the importance for each search keyword from the specified search keyword and the operation content and the added points to be added to the importance for each search keyword And a document-to-word evaluation score update step for updating the action-to-evaluation base score for each Web page based on the operation content and the added points of the search keyword. A web search method is provided.

  The present invention is a program for executing the web search method described above, and may be a computer-readable recording medium storing the program.

  According to the present invention, there is an advantage that the search result is close to the target result of the user in the order of presentation to the user.

It is a functional block diagram which shows the example of 1 structure of the web search system by embodiment of this invention. It is a figure which shows the example of 1 structure of a document versus word evaluation score database. It is a flowchart figure which shows the flow of a search process. It is a figure which shows the process until it changes from step S309 of FIG. 3A to step S302. It is a figure which shows the detail of a process of step S303 (search result list generation process) of FIG. 3A. It is a figure which shows the detail of the process of step S307 (search keyword weight calculation) of FIG. 3A. It is a figure which shows the detail of the process of step S308 (calculation evaluation addition point) of FIG. 3A. It is a figure which shows the detail of a process of step S309 (update of the document versus word evaluation score of a web page) of FIG. 3A. It is a figure which shows the action versus evaluation base score which shows the correspondence of operation content (action) and a score point. It is a figure which matches the keyword which lowers keyword weight, and a subtraction value.

  The inventor gives evaluation scores of “common words that are not very important for Web pages” and “words that are important keys for Web pages” in an efficient Web search result presentation system and method based on user evaluation feedback. If treated equally, “Web Page A Evaluation Score Total”> “Web Page B Evaluation Score Total”, and the user is presented with Web Page A with priority even though he wants to know the programming language B specifications. As a result, there was a problem that the quality of search results deteriorated.

  Hereinafter, a Web search technique according to an embodiment of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a functional block diagram showing a configuration example of a Web search system according to an embodiment of the present invention. The Web search system according to the present embodiment includes a host system (search device) 110 and a document-to-word evaluation score database 119 connected thereto. The host system 110 is connected to a network 120 such as the Internet. The network 120 stores a user terminal 100 that uses the Web search system according to the present embodiment and a Web page document to be searched. A server group 130 is connected.

  The host system 110 includes a document collection unit 111 that collects Web page documents from the server group 130, a document-to-word evaluation score initialization unit 112 that initializes data in the document-to-word evaluation score database 119, and a user terminal 100. A search request receiving unit 113 that receives a search request from the keyword, a keyword importance determining unit 114 that determines the importance of the keyword from the search request keyword received by the search request receiving unit 113, and the search request keyword received by the search request receiving unit 113 A search result list generation unit 115 that determines a page rank from information in the document versus word evaluation score database 119 and generates a search result list, a search result list transmission unit 116 that transmits the search result list to the user terminal 100, and a user terminal The user evaluation result of the search result list transmitted to 100 is received. The user evaluation receiving unit 117 includes a document-to-word evaluation score updating unit 118 that updates data in the document-to-word evaluation score database 119 using the above-described evaluation result and the keyword importance determined by the keyword importance determining unit 114. ing.

  The document-to-word evaluation score database 119 stores data having a structure as illustrated in FIG. As indicated by reference numeral 200 in FIG. 2A, a plurality of pieces of web page information are stored for each web page address. Each piece of Web page information is composed of a plurality of pieces of information as indicated by reference numerals 210 to 241. The web page address 210 is the address of the document of the web page. Each keyword included in the document at this address has an evaluation score. In the example of FIG. 2, the evaluation score for keyword 1 (220) in the document is the evaluation score 1 (221), the evaluation score for keyword 2 (230) is the evaluation score 2 (231), and the evaluation score for keyword 3 (240) is Evaluation score 3 (241),...

  FIG. 3 is a flowchart showing the flow of search processing. FIG. 3A is a flowchart showing the overall processing flow.

  First, FIG. 3A will be described. A step S301 collects Web page documents from the server group 130 connected to the network 120 using the document collection unit 111, and extracts, for example, all keywords for each Web page using the document-to-word evaluation score initialization unit 112. Then, the evaluation score is initialized to the document of the Web page address and stored in the document-to-word evaluation score database 119 shown in FIG. In step S302, the search request receiving unit 113 is used to receive a search request from the user. In step S303, the search result list generation unit 115 determines the Web page rank based on the Web page information in the document-to-word evaluation score database 119 and the search keyword of the search request received in step S302, and generates a search result list. Details of the processing in step S303 will be described later. In step S304, the result list generated in step S303 is transmitted to the user terminal 100. In step S <b> 305, the user evaluation receiving unit 117 receives what action (action X) the user has taken on the web page at which address (address A) in the presented result list.

  Thereafter, steps S306 to S309 are score update processing performed by the document-to-word evaluation score update unit 118. In step S306, as shown in FIG. 4, an evaluation-based score (hereinafter referred to as “ Evaluation base score B ”) is determined. For example, if the action (action X) taken by the user is “print page (403)”, the evaluation base score B is “evaluation base score 2 (404)”. In step S307, the keyword importance level determination unit 114 determines the keyword importance level (weight value) for each search keyword. Details of the processing in step S307 will be described later.

  In step S308, the evaluation score added value for each keyword is determined by associating the evaluation base score determined in step S306 with the importance for each keyword determined in step S307 by a process such as multiplication. Details of this processing will be described later. In step S309, the addition point determined in step S308 is added to the evaluation score for each keyword with respect to the record in which the Web page address in the document-to-word evaluation score database 119 matches the address A. Details will be described later. Thereafter, the process returns to the search request receiving process in step S302. By repeating these steps S302 to S309, the search result list can be gradually improved.

  FIG. 3B is a diagram illustrating processing from the transition from step S309 to step S302. The following processing is also performed at appropriate times during the loop returning to step S302. First, in step SA1, it is determined whether the text of the Web page is added to the server. If Yes, the process proceeds to step SA2, and the initialization process of step S301 is newly added to the server group 130 at regular intervals. In step SA2, initialization is performed in step S301 at regular intervals with respect to the web page. In step SA3, the newly added web page document is document-to-word evaluated. It adds to the score database 119. In step SA4, if there is a deleted Web page document, it is deleted from the document-to-word evaluation score database 119, and the process proceeds to step S302.

  Next, details of step S303 (search result list generation processing) will be described with reference to FIG. 3C. In step S303-1, a Web page document including all the search keywords received in step S302 is selected from the document-to-word evaluation score database 119, and the total evaluation score is calculated. For example, it is assumed that the search keywords are “keyword 1”, “keyword 2”, and “keyword 3” from the top. When the Web page information including all of these keywords is from 210 to 241 in FIG. 2B, the total evaluation score of this Web page is “evaluation score 1 (221)” + “evaluation” Score 2 (231) "+" Evaluation score 3 (241) ". Such a calculation is performed for Web page information including all of the keywords “keyword 1”, “keyword 2”, and “keyword 3”, and a list in which the Web page addresses 210 are arranged in descending order of the total evaluation score is generated. In step 303-2, calculation is performed for the Web page including all keywords. If the calculation has been completed (YES), a list is created in descending order of the total evaluation score in step 303-3, and the list is calculated in step S304. Send.

  Next, details of step S307 (search keyword weight calculation) will be described with reference to FIG. 3D. If the search keywords received in step S302 are “keyword 1”, “keyword 2”,..., “Keyword n” from the top, the weights of the obtained keywords are “weight W1,” “weight W2,”.

First, based on the assumption that the keyword specified at the top as a general usage method is an important keyword for the user of the search system, a basic value that satisfies W1>W2>. S307-1). For example, weighting is performed for each keyword as follows.

  Next, in “keyword 1” “keyword 2”... “Keyword n”, based on the table associating the keyword weighting and the subtraction value as shown in FIG. (Step S307-2), and if there is a matching keyword (Yes in step S307-3), a subtraction process is performed to reduce the weight determined above by the “importance subtraction value” that matches the keyword. (Step S307-4). For example, when “keyword 2” matches “keyword B (503)”, “importance subtraction value B” is subtracted from “weight W2”. At this time, the value of “weight W2” is adjusted so as not to fall below the minimum value (> 0) of the weight. The weights of keywords that do not match the keywords in the table of FIG. 5 are left as they are (No in step S307-3).

Next, with reference to FIG. 3E, the details of step S308 (calculate evaluation addition points) will be described. In step S308-1, it is determined whether or not there is an evaluation addition term that matches the keyword. In the case of Yes, addition (for each keyword) is performed by evaluation addition in step S308-2. That is, the evaluation addition score A for each keyword is determined by multiplying the evaluation base score B determined in step S306 by the weight W determined in step S307. For example, the addition points are obtained for each keyword as follows.

In more detail, it is as follows.
“Evaluation addition point A1” = “Evaluation base score B” × “Weight W1” of “Keyword 1”
“Evaluation addition point A2” = “Evaluation base score B” × “Weight W2” of “Keyword 2”
:
“Evaluation addition point An” = “Evaluation base score B” × “Weight Wn” of “Keyword n”

  With reference to FIG. 3F, details of step S309 (update of document-to-word evaluation score of Web page) will be described. The evaluation addition score determined in step S308 is reflected in the information of the document-to-word evaluation score database 119 (step S309-1). For the Web page information that matches the address A acquired in step S305 and the Web page address (210) (Yes in step S309-2), the evaluation score of the above-mentioned “keyword 1” to “keyword n” is “evaluation score A1”. -"Evaluation score An" is added (step S309-3), and the document-to-word evaluation score database DB119 is updated (step S309-4).

  As described above, the search result presentation system according to the present embodiment has the following advantages.

  Analyze search request keywords, calculate the importance of each keyword, and use the importance as a weight to the keyword for each Web page when adding the evaluation score of the database of document score vs. word evaluation score of the Web page This increases the accuracy of the Web page vs. keyword evaluation score data. In addition, the page rank calculation result for determining the order in which the search results are presented to the user is closer to the target result of the user.

  In the above-described embodiment, the configuration and the like illustrated in the accompanying drawings are not limited to these, and can be changed as appropriate within the scope of the effects of the present invention. In addition, various modifications can be made without departing from the scope of the object of the present invention. Each component of the present invention can be arbitrarily selected, and an invention having a selected configuration is also included in the present invention.

  In addition, a program for realizing the functions described in the present embodiment is recorded on a computer-readable recording medium, and the program recorded on the recording medium is read into a computer system and executed to execute processing of each unit. May be performed. The “computer system” here includes an OS and hardware such as peripheral devices.

  Further, the “computer system” includes a homepage providing environment (or display environment) if a WWW system is used.

  The present invention can be used in a search system.

DESCRIPTION OF SYMBOLS 100 ... Search system user terminal, 110 ... Search host system, 111 ... Web page document collection part, 112 ... Web page document versus word evaluation score initialization part, 113 ... Search request reception part, 114 ... Search keyword importance determination part 115 ... Search result list generating unit 116 ... Search result list transmitting unit 117 ... User evaluation receiving unit 118 ... Web page document-to-word evaluation score update unit 119 ... Web page document-to-word evaluation score 120, network, 130, a server group storing web page documents connected to the network.

Claims (5)

  Collecting Web pages and initializing document-to-word evaluation scores for Web pages, search-request receiving unit for receiving specified search keywords, and search keywords from specified search keywords Referring to the keyword importance level determination unit that determines the importance level of the operation, and the storage unit that stores the action vs. the evaluation-based score indicating the correspondence between the operation content and the added point added to the importance level for each search keyword, A web search apparatus, comprising: a document-to-word evaluation score update unit that updates the action-based evaluation base score for each Web page based on content and the added points of the search keyword. The document-to-word evaluation score update unit
2. The web according to claim 1, wherein for each web page, the action vs. evaluation base score for each web page is updated based on a subtraction point indicating a search keyword from which the weight of the search keyword is subtracted. Search device. The document-to-word evaluation score update unit
Update to the document-to-word evaluation score for each Web page at a certain time interval, and if there is a deleted Web page document, continue to delete it from the document-to-word evaluation score The web search device according to claim 1 or 2. Document-to-word evaluation scoring step for collecting web pages and initializing document-to-word evaluation scores on the web pages;
A search request receiving step for receiving a specified search keyword;
A keyword importance determination step for determining the importance of each search keyword from the specified search keywords;
With reference to the storage unit storing action vs. evaluation-based score indicating the correspondence between the operation content and the point added to the importance for each search keyword, based on the operation content and the point added to the search keyword, A web search method comprising: a document vs. word evaluation score update step for updating the action vs. evaluation base score for each Web page.   A program for causing a computer to execute the web search method according to claim 4.

Images