JP7109764B2 - Evaluation device and evaluation method - Google Patents

Description

The present invention relates to a technology for evaluating web pages, and in particular, when a web page has a link relationship with another web page with more relevant content, it is evaluated more highly. The present invention relates to a technique for allowing a search engine or the like to highly evaluate web pages having a more preferable link relationship, thereby making it easier to preferentially display them to the user.

Today, the search engines of major operating companies, represented by Google (registered trademark), are used for web pages that take a long time to be viewed by visitors to web pages through searches (hereafter referred to as search visitors), and for pages that are viewed per visit. It is empirically known among marketers that a website with a large number of pages viewed (hereinafter referred to as the number of pages viewed) tends to be ranked higher for the relevant search term. For this reason, search engine operators use access analysis tools, etc., to measure or collect trends in the behavior of search visitors on each website for each search term, and use them as part of the index for determining search rankings. It is presumed that

Under such circumstances, for website creators, etc., to increase the number of web pages viewed and the time spent browsing web pages by search visitors in order to achieve high ranking in search results for specific search terms on search engines. , it is desirable to make the description content of the web page suitable for the search term. Further, when the search visitor transitions from the web page to another linked web page, it is further desirable that the linked web page also maintain some kind of relevance. Specifically, the content of the first web page viewed by the search visitor for the first time and the content of the second web page to which the search visitor clicks a link from that page are measured and compared using natural language processing technology. In this case, it is desirable that both contents are confirmed to have a higher degree of relevance, and that both contents are more suitable for the search term. On the other hand, it is not preferable that the content of both pages is poorly related, and that the search visitor is led to a linked page that does not match the search purpose. In this way, various techniques for improving search results have been proposed from the viewpoint of considering the link relationship between web pages from the viewpoint of the convenience of search visitors.

For example, Patent Literature 1 discloses a technique for processing web pages of websites in order to improve search results. More specifically, the relationship between the identified terms and the second web page such that the second web page will have higher search results for search queries that include the identified search query terms. As an action to save the rating, a process to increase the page rank component of the search score is executed.

Japanese Patent Publication No. 2009-532766

However, the invention disclosed in U.S. Pat. No. 6,200,000 performs a second test and, if it is determined that the search term may be suitable for a second web page, passes the search term to the second web page. The second test is to determine the page type of the second web page and exclude it from the relationship if it is a web page of a given type. is. In other words, the technique disclosed in the document increases the page rank component of the search score in order to associate the search term with the second web page when a predetermined test is cleared. It merely reflects the relevance of the web pages in the search score, and does not reflect the relevance between the first web page and the second web page in various indices of the first web page. Therefore, when the search word is "sofa bed", the relevance between the "sofa bed" and the second web page is measured, and when the predetermined test is cleared, the relevance with the search word is evaluated as the second web page. It is only reflected in the search score of the web page.

In general, no matter how high the relevance between a search term and the contents of the web page visited by the searcher by clicking on the search result is, the search visitor may search the web page to obtain more detailed or related information. It is often the case that a transition to another web page, ie, a second web page, is made by following an in-site menu or an external link described on the page. In such a case, if the transition to the linked web page is completely unrelated to all or part of the content described on the first web page that is the link source, in order to achieve the purpose of the search Then, the user has to return to the first web page again or search another web page again using a search engine or the like. That is, the lower the relevance between the search term initially used by the search visitor and the content of the first web page and the content of the second web page, the lower the probability that the search visitor can achieve the search purpose. It is believed that there is.

Therefore, in discovering and evaluating web pages with appropriate content for a given search term, not only the content of the first web page that the search visitor first visits, but also the content of that web page and its links. It is considered desirable to evaluate the contents of the second web page, which is related, and add this to the evaluation of the first web page. rice field.

The present invention has been made in view of such problems. , to provide a technique for evaluating a web page having a more favorable link relationship.

In order to solve the above-mentioned problems, the present invention provides a designating means for designating a web page to be evaluated, a search term likely to be used when retrieving the web page to be evaluated, and a web page to be evaluated. , a direct link page that is a web page that is directly linked to the web page to be evaluated, and an indirect link page that is a web page that is linked to the web page to be evaluated via one or more other web pages. and identifying means for analyzing the web page data of each and identifying the link relationship between the web pages; and a correction means for correcting the index set for each web page by the setting means according to the link relationship specified by the specifying means. An evaluation device characterized by:

In the present invention, the index set for each of the web page to be evaluated, the direct link page, and the indirect link page is the index of the web page to be evaluated and the web page directly or indirectly linked to the web page. Corrected according to the link relationship. Therefore, according to the present invention, it is possible to evaluate the quality of a web page including the link relationship with other web pages linked to the web page to be evaluated.

In a more preferred embodiment of the evaluation apparatus, the designation means designates the number of layers N of link relationships among the web page to be evaluated, the direct link page, and the indirect link page. In general, as the number of layers N increases, the number of indirect link pages increases, and the processing load on each of the identifying means, the setting means, and the correcting means increases. According to this aspect, it is possible to specify and evaluate a suitable number of layers N in consideration of the processing load.

As a specific example of the index set by the above setting means, even if it is a value based on the subjectivity of the subject who compared both the posted information on the web page and the search term that would be used when searching for the web page Alternatively, it may be a value obtained by objectively analyzing posted information on a web page.

Further, in order to solve the above-described problems, the present invention provides a designating means for designating a web page to be evaluated, the web page to be evaluated, and a direct link web page directly linked to the web page to be evaluated. Analyzing web page data of each of a page and an indirect link page, which is a web page linked to the web page to be evaluated via one or more other web pages, to identify the link relationship between the web pages. specifying means; and calculating means for calculating an index indicating the quality of the link relationship according to the appearance frequency of each of a plurality of predetermined types of link patterns in the link relationship specified by the specifying means. An evaluation device characterized by: According to the present invention, it is possible to evaluate the quality of the link relationship between an evaluation target page and another web page directly or indirectly linked to the evaluation target page.

Further, in order to solve the above-described problems, the present invention provides a designating step of designating a web page to be evaluated and a search term likely to be used when retrieving the web page to be evaluated; a page, a direct link page that is a web page that is directly linked to the web page to be evaluated, and an indirect link that is a web page that is linked to the web page to be evaluated via one or more other web pages. an identification step of analyzing web page data of each page and identifying a link relationship between web pages; a setting step of setting, for each web page, an index indicating relevance to a word; and a correction of correcting the index set for each web page in the setting step according to the link relationship identified in the identification step. and an evaluation method characterized by comprising: According to the present invention, it is also possible to evaluate the power of content including the link relationship with other web pages linked to the web page to be evaluated.

Further, in order to solve the above-described problems, the present invention provides a designating step of designating a web page to be evaluated, the web page to be evaluated, and a direct link web page directly linked to the web page to be evaluated. Analyzing web page data of each of a page and an indirect link page, which is a web page linked to the web page to be evaluated via one or more other web pages, to identify the link relationship between the web pages. and a calculating step of calculating an index indicating the quality of the link relationship according to the appearance frequency of each of a plurality of predetermined types of link patterns in the link relationship specified in the specifying step. and an evaluation method characterized by: According to the present invention, it is also possible to evaluate the quality of the link relationship between the web page to be evaluated and other web pages linked to the web page.

According to the present invention, by using natural language processing technology to evaluate the relevance of the contents of a web page to be evaluated and other web pages that have a link relationship therewith, a web page having a more favorable link relationship can be ranked higher. We can provide technology for evaluation.

It is a figure showing the whole composition of evaluation system 1A containing evaluation device 20A by a 1st embodiment of the present invention. It is a flowchart which shows the flow of the evaluation method which CPU22 of 20 A of evaluation apparatuses performs according to 26 A of evaluation programs. It is a figure for demonstrating the designation|designated process of this embodiment. It is a figure for demonstrating the correction|amendment process of this embodiment. It is a figure which shows the example of a display of the evaluation result by the evaluation apparatus of this embodiment. It is a figure which shows the whole structure of the evaluation system 1B containing the evaluation apparatus 20B by 2nd Embodiment of this invention. 3 is a flow chart showing the flow of an evaluation method executed by a CPU 22 of an evaluation device 20B according to an evaluation program 26B; It is a figure for demonstrating the designation|designated process of this embodiment. It is a figure for demonstrating the calculation process of this embodiment.

Embodiments of the present invention will be described below with reference to the drawings.
(A: First Embodiment)
FIG. 1 is a diagram showing the overall configuration of an evaluation system 1A including an evaluation device 20A that is one embodiment of the present invention. The evaluation system 1A determines the relevance (conceptually including affinity, which means a more compatible relationship, the same applies hereinafter) between a web page and a search term that will be used when searching the web page for that web page, Contents of other web pages linked to the web page and link relationships are also evaluated. As shown in FIG. 1, the evaluation system 1A has a user terminal 10 and an evaluation device 20A. The user terminal 10 and the evaluation device 20A are connected via a network 90 such as the Internet.

The user terminal 10 is a personal computer. A unique ID and password are assigned to the user of the user terminal 10 . The user accesses the evaluation device 20A from his/her own user terminal 10, performs an authentication procedure, and uses the service.

Although not shown in FIG. 1, the network 90 is connected to one or more search engine server devices that serve to provide search engine services. The search engine service receives an HTTP (Hypertext Transfer Protocol) request (search request) containing a search term from a computer connected to the network 90, and transmits the title, URL, and snippet of the website containing the search term. A network service that returns search results arranged in order from the highest search ranking.

The evaluation device 20A has a communication interface 21, a CPU (Central Processing Unit) 22, a RAM (Random Access Memory) 23, a ROM (Read Only Memory) 24, and a hard disk 25, as shown in FIG. Communication interface 21 transmits and receives data to and from devices connected to network 90 . The CPU 22 executes various programs stored in the ROM 24 and the hard disk 25 while using the RAM 23 as a work area. The ROM 24 stores an IPL (Initial Program Loader) and the like. The hard disk 25 stores an evaluation program 26A that causes the CPU 22 to implement functions unique to this embodiment.

Next, the operation of this embodiment will be described. The CPU 22 executes IPL when the power of the evaluation device 20A is turned on, reads the evaluation program 26A from the hard disk 15 to the RAM 23 according to this IPL, and starts its execution. FIG. 2 is a flow chart showing the flow of the evaluation method executed by the CPU 22 according to the evaluation program 26A. By executing the evaluation program 26A, the CPU 22 performs designation means for executing the designation process (step SA100), identification means for executing the identification process (step SA110), setting means for executing the setting process (step SA120), and correction It functions as correction means that executes the process (step SA130).

In the designation process of step SA100, as shown in FIG. 3, the CPU 22 transmits web page data describing the input screen relationship in HTML (HyperText Markup Language) to the user terminal 10 that has completed the authentication procedure. When receiving the web page data, the user terminal 10 presents an input screen to the user. The input screen displays a character string "Specify a web page to be evaluated". Below that, there is a URL input field 51 . In the URL input field 51, the user of the user terminal 10 enters the URL of a web page whose relevance to the search term is to be evaluated (hereinafter referred to as the page to be evaluated: for example, the web page of the site operated by the user). is entered.

As shown in FIG. 3, below the URL input field 51, a character string such as "Specify a search term" is displayed. Below that, there is a search term input field 52. Below 52 is a send button 53 . In the search term input field 52, the user of the user terminal 10 inputs a search term that will be used when searching the web for the page to be evaluated.

After completing the information input in the URL input field 51 and the search word input field 52 , the user presses the send button 53 . When the send button 53 is pressed, the user terminal 10 sends the information entered in the URL input field 51 and the search term input field 52 to the evaluation device 20A as input information. When receiving the input information from the user terminal 10 via the network 90 , the CPU 22 writes the received input information to the RAM 23 .

Next, CPU 22 executes a specific process (step SA110). In this identification process, the CPU 22 determines the evaluation target page, the direct link page which is a web page directly linked to the evaluation target page, and the web pages linked to the evaluation target page via one or more other web pages. A link relation between each web page of an indirect link page, which is a page, is specified. Here, the web page directly linked to the evaluation target page refers to a web page in which the URL of the web page is embedded as a hyperlink in the evaluation target page. A web page indirectly linked to the page to be evaluated is a web page whose URL is not embedded in the page to be evaluated. The indirect link page includes a web page in which the URL of the web page is embedded.

More specifically, the CPU 22 first accesses the URL containing the input information written in the RAM 23 in step SA100, and retrieves the web page data indicated by the URL (that is, the web page data of the page to be evaluated). It acquires and writes in RAM23. The CPU 22 obtains the URL of the directly linked page by extracting the URL embedded in the web page data as a hyperlink from the web page data. The CPU 22 associates the URL of the page to be evaluated with the URL of the direct link page to generate link relationship data indicating the link relationship between the two web pages, and writes the data to the RAM 23 . Next, the CPU 22 accesses the URL of the directly linked page obtained in the manner described above, obtains the web page data of the directly linked page, and writes it to the RAM 23 .

Next, the CPU 22 acquires the URL of the indirect link page directly linked to the direct link page by extracting the URL embedded as a hyperlink in the web page data of the direct link page. The CPU 22 associates the URL of the direct link page with the URL of the indirect link page to generate link relationship data indicating the link relationship between the two web pages, and writes the link relationship data to the RAM 23 .

The CPU 22 accesses the URL of the indirect link page written in the RAM 23, acquires the web page data of the indirect link page, writes it in the RAM 23, and extracts the URL embedded as a hyperlink in the web page data of the indirect land page. By doing so, the URL of another indirect link page that is directly linked to the indirect link page is obtained, and the URL of the indirect link page and the URL of the other indirect link page are associated to establish the link relationship between both web pages. The process of generating the link relation data shown and writing it to the RAM 23 is repeated (N-1) times or until the terminal indirect link page is reached. Here, N is a natural number of 1 or more predetermined in the evaluation program 26A, and represents the number of layers of link relationships among the web pages of the evaluation target page, the direct link page, and the indirect link page.

Next, CPU 22 executes a setting process (step SA120). In this setting process, for each of the evaluation target page, the direct link page, and the indirect link page, the CPU 22 sets the index indicating the relevance to the search term or the evaluation target page specified in the input process at step SA100 to the web page. set for each In the setting process of this embodiment, the CPU 22 sets the evaluation target page, the direct link page, and the indirect link page as the analysis target web pages in Japanese Patent No. 6164436, and calculates according to the algorithm disclosed in the Japanese Patent No. 6164436. The evaluation score regarding the degree of portal that is used is set as the above index. In the present embodiment, the evaluation score related to the degree of portal is set as the above index. , or a value calculated according to another algorithm.

In the correction process (step SA130) subsequent to the setting process (step SA120), the CPU 22 converts the index set for each web page in the setting process (step SA120) into the link relationship specified in the specifying process (step SA110). correct accordingly. The processing contents of this correction processing will be described below.

For example, the URLs of the direct link page P2 and the direct link page P3 are embedded as hyperlinks in the evaluation target page P1, and the URLs of the direct link page P3 and the indirect link page P4 are embedded as hyperlinks in the direct link page P2. It is assumed that the URL of the direct link page P2 is embedded as a hyperlink in the direct link page P3. In this case, the link relationship among the evaluation target page P1, the direct link page P2, the direct link page P3, and the indirect link page P4 is represented as shown in FIG. In FIG. 4, each web page is represented by a circle, and links between web pages are represented by arrows. Numerical values (480, 400, 320, and 130, respectively) shown in circles in FIG. 4 are index values set in the setting process for the web pages corresponding to the circles. In the example shown in FIG. 4, the domains of the direct link page P2, the direct link page P3, and the indirect link page P4 are the same, and only the evaluation target page P1 has a different domain. In FIG. 4, circles indicated by symbols P2, P3, and P4 are given the same hatching, and the circle indicated by symbol P1 is not hatched, which indicates the relationship between the above domains. .

Using the diagonal matrix C (see formula (1)) in which the index set in the setting process (step SA120) for each web page is arranged on the diagonal elements and the matrix A representing the correction according to the link relationship , the corrected index matrix C′ is represented by the following equation (2). The matrix C, the matrix C′, and the row i column j row component (i≠j) in the matrix A reflect the link relationship between the web page Pi and the web page Pj. In the setting process (step SA120), the indices are set without considering the link relationship, so all the elements other than the diagonal elements in the matrix C are zero.

The matrix A representing the correction according to the link relationship includes a matrix L _oe representing the external outbound link, a matrix L _oi representing the internal outbound link, a matrix L _ie representing the external inbound link, a matrix L ii representing the internal inbound link, and a matrix L _ii representing the internal inbound link. A scalar w _oe , the weighting factor for the outbound link _; a scalar w _oi , the weighting factor for the internal outbound link; a scalar w _ie , the weighting factor for the external inbound link; is represented by the following formula (3).

An external outbound link is a link in the direction from the link source web page to the link destination web page where the domains of the link source web page and the link destination web page are different. The i column j row element of the matrix _Loe representing the external outbound link is set to 1 if the web page Pi and the web page Pj have different domains and there is a link in the direction from the web page Pi to the web page Pj. , otherwise 0 is set. An external inbound link is a link in which the domains of the link source web page and the link destination web page are different, and the direction is opposite to that of the external outbound link. The matrix L _ie representing the external inbound link is the transpose of the matrix L _oe representing the external outbound link. In the link relationship shown in FIG. 4, the links are directed from the evaluation target page P1 to each of the direct link page P2 and the direct link page P3 having different domains from the evaluation target page P1. Therefore, the matrix L _oe representing the external outbound link is represented by the following equation (4), and the matrix L _ie representing the external inbound link is represented by the following equation (5).

An internal outbound link is a link in which the domains of the link source web page and the link destination web page are the same, and the link is directed from the link source web page to the link destination web page. The i column j row element of the matrix L _oi representing the internal outbound link contains 1 when the web page Pi and the web page Pj have the same domain and there is a link in the direction from the web page Pi to the web page Pj. set, 0 otherwise. An internal inbound link is a link in which the domains of the link source web page and the link destination web page are the same, and the direction is opposite to that of the internal outbound link. The matrix L _ii representing the internal inbound link is the transpose of the matrix L _oi representing the external outbound link. In the link relationship shown in FIG. 4, the direct link page P2 is linked to the direct link page P3 and the indirect link page P4 belonging to the same domain, and the direct link page P3 is linked to the direct link page P2. there is Therefore, the matrix L _oi representing the internal outbound link is represented by the following equation (6), and the matrix L _ii representing the internal inbound link is represented by the following equation (7).

In relation to web page links, it is considered that a link from a web page with more links is more highly evaluated. In order to enable such evaluation, the present embodiment expands the formula (2) and first defines the recurrence formula of the following formula (8). Matrix A _n (n≧1) in equation (8) is a matrix representing a correction according to the link relationship of the n-th layer when the evaluation target page is the 0th layer in the link relationship of the web page, and the matrix L is represented by the following equation (9). In this case, if the scalar that is the weighting coefficient of the link relation in the _n -th layer is wn, the above formula (2) is extended to the following formula (10).

In the correction process (step SA130) of the present embodiment, the CPU 22 sets the index matrix C set for each web page in the setting process (step SA120) and the link relationship specified in the specification process (step SA110). Using the corresponding matrix An (n = 0 to N), the matrix C' related to the index after correction is calculated according to the above formula (10), and the matrix C' is used by the user terminal 10 that specified the evaluation target page. Reply to

In the first term on the right side of equation (10), the relevance between the content of each web page of the evaluation target page, the direct link page, and the indirect link page and the search term that will be used when searching for the evaluation target page is reflected, and the link relationship between each web page is reflected in the second term on the right side. Therefore, the user of the user terminal 10 who specified the page to be evaluated refers to the index C′ to determine the relationship between the web page to be evaluated and the search term likely to be used when searching for the web page. With respect to the nature of a web page, it is possible to obtain an evaluation result that takes into account the link relationship with other web pages linked to the web page to be evaluated and the content of the other web pages. As described above, according to the present embodiment, the relevance between a web page to be evaluated and a search term that is likely to be used when searching for the web page is determined by other search terms linked to the web page to be evaluated. It becomes possible to evaluate the link relationship with the web page and the content of the other web page.

A display example of an actual evaluation result is as shown in FIG. 5, for example. Here, as an example, the results of searching for the search keyword "sofa bed price" are shown, and the overall evaluation of the web page to be evaluated is shown, and the breakdown thereof is 1) evaluation of the web page to be evaluated alone, 2) Link-related web page evaluations are indicated by scores. In addition, in link-related web page evaluation, it is possible to check the actual link destination URL and the evaluation score of each link destination. In this way, it is possible to evaluate the relevance between a web page to be evaluated and a search term that is likely to be used when retrieving the web page, taking into consideration the link relationship.

(B: Second Embodiment)
Next, a second embodiment of the invention will be described.
FIG. 6 is a diagram showing the overall configuration of an evaluation system 1B including an evaluation device 20B according to the second embodiment of the invention. The evaluation system 1B evaluates the link relationship between the evaluation target page and other web pages directly or indirectly linked to the evaluation target page. As is clear from comparing FIG. 1 and FIG. 6, the evaluation system 1B differs from the evaluation system 1A in that it has an evaluation device 20B instead of the evaluation device 20A. The evaluation device 20B has the same hardware configuration as the evaluation device 20A, but differs from the evaluation device 20A in that the evaluation program 26B is stored in the hard disk 25 instead of the evaluation program 26A.

Also in this embodiment, the CPU 22 executes IPL when the evaluation device 20B is powered on, reads the evaluation program 26B from the hard disk 15 to the RAM 23 according to this IPL, and starts executing it. FIG. 7 is a flow chart showing the flow of the evaluation method executed by the CPU 22 of the evaluation device 20B according to the evaluation program 26B. As is clear from comparing FIG. 2 and FIG. 7, the evaluation method in this embodiment includes a designation process (step SB100) instead of the designation process (step SA100), and a setting process (step SA120). ), and the calculation process (step SB120) instead of the correction process (step SA130). That is, by executing the evaluation program 26B, the CPU 22 of the evaluation device 20B performs designation means for executing the designation processing (step SB100), identification means for executing the identification processing (step SA110), and calculation processing (step SB120). It functions as a calculation means to be executed. Hereinafter, the specification processing (step SB100) and the calculation processing (step SB120), which are different from the first embodiment, will be mainly described.

In the designation process of step SB100, the CPU 22 transmits web page data describing the relationship of the input screens in HTML to the user terminal 10 that has completed the authentication procedure, as shown in FIG. When receiving the web page data, the user terminal 10 presents an input screen to the user. On this input screen, a character string "Specify a web page to be evaluated" is displayed. Below that, there is a URL input field 51 . In the URL input field 51, the user of the user terminal 10 inputs the URL of the web page to be evaluated for the quality of the link relationship. Below the URL input field 51 is a send button 53 . As is clear from a comparison between FIG. 3 and FIG. 8, the input screen of this embodiment does not include the character string "Please enter a search term", and the search term input field 52 is also Not provided. The input screen of this embodiment differs from the input screen of the first embodiment (see FIG. 3) in these respects. This is because, in this embodiment, the link relationship between the evaluation target page and other web pages directly or indirectly linked to the evaluation target page is evaluated, and there is no need to specify a search term.

After completing the information input in the URL input field 51 , the user presses the send button 53 . When the send button 53 is pressed, the user terminal 10 sends the information input in the URL input field 51 (that is, the URL of the page to be evaluated) as input information to the evaluation device 20B. When receiving the input information from the user terminal 10 via the network 90 , the CPU 22 writes the received input information to the RAM 23 .

The processing contents of the specifying process (step SA100) executed subsequent to the specifying process (step SB100) are the same as those of the specifying process in the first embodiment. In other words, in this specifying process (step SA110), the CPU 22 obtains web page data for each of the evaluation target page, the direct link page, and the indirect link page, analyzes the web page data, and determines the link relationship of each web page. is specified, and link relation data indicating the link relation is written in the RAM 23 .

In the calculation process (step SB120) subsequent to the identification process (step SA110), the CPU 22 calculates the frequency of appearance of each of a plurality of predetermined link patterns in the link relationship identified in the identification process. An index S indicating the quality of the link relationship is calculated according to the appearance frequency, and the index S is returned to the user terminal 10 that specified the evaluation target page. In this embodiment, the plurality of types of predetermined link patterns are link patterns PT01 of the other two nodes when one particular node included in the triad is the parent node (hatched in FIG. 9). to PT33 (see FIG. 9). To each of the link patterns PT01 to PT33 shown in FIG. 9, a larger evaluation value is assigned in advance as the number of links included in the link pattern (indicated by arrows in FIG. 9) is larger. This is because a link relationship with a larger number of links is considered preferable.

Details of the processing contents of the calculation processing (step SB120) are as follows. The CPU 22 first sets the index S to an initial value (for example, 0). Next, the CPU 22 selects the most upstream web page as the page to be evaluated in the link relationship represented by the link relationship data stored in the RAM 23, and selects web pages one by one from the upstream side as the parent node. It is determined which of the link patterns PT01 to PT33 the link relationship between the parent node and the web page directly linked to the one downstream side corresponds to, and the evaluation value of the link pattern determined to correspond is added to the index S. do.

When the selected web page is the evaluation target page, the link relationship between the evaluation target page and the direct link page corresponds to any of the link patterns PT01 to PT33 shown in FIG. 9, and the evaluation value of the corresponding link pattern is It is added to index S. For example, if there is one direct link page, the evaluation value of any of the link patterns PT01 to PT03 in FIG. 9 is added to the index S. If there are two direct link pages, the link patterns PT04 to PT04 in FIG. Any evaluation value of PT09, PT13 to PT18, PT22 to PT24, and PT28 to PT33 is added to the index S. If there are three or more directly linked pages, the evaluation value of any of the link patterns PT01 to PT03 and the evaluation value of any of the link patterns PT04 to PT09, PT13 to PT18, PT22 to PT24, and PT28 to PT33 is the index S. is added to Similarly, the link relationship between the direct link page and the indirect link page that is directly linked to the directly linked page, and the link relationship between the indirect link page and the indirect link page that is directly linked to the indirect link page. is added to the index S.

According to this embodiment, the index S indicating the quality of the link relationship between the page to be evaluated and another web page directly or indirectly linked to the page to be evaluated is used by the designated source of the page to be evaluated. By referring to the indicator S, which is sent back to the user terminal 10, the user of the user terminal 10 can grasp the quality of the link relationship.

(C: modification)
The first and second embodiments of the present invention have been described above. However, of course, the following modifications may be added to these embodiments.
(1) In the first embodiment, the number of layers N in the link relationship among the web pages of the evaluation target page, the direct link page, and the indirect link page was fixed. can be specified. Specifically, the input screen shown in FIG. may be used as a trigger to cause the user terminal 10 to execute. In general, the greater the number of layers N, the heavier the processing loads of the identification process (step SA110), the setting process (step SA120), and the correction process (step SA130). It is possible to specify and evaluate a suitable number of layers N in consideration of the processing loads of the identification process (step SA110), the setting process (step SA120), and the correction process (step SA130).

(2) The matrix L in the first embodiment is a matrix obtained by weighted addition of the matrix L _oe , the matrix L _ie , the matrix L _oi , and the matrix L _ii . However, when evaluating the relevance to a search term only for links with different domains between the link source web page and the link destination web page, the matrix L is a weighted addition of the matrix L _oe and the matrix L _ie . It's good if it's a matrix that does. Similarly, when evaluating the relevance to a search term only for links in which the domain of the link source web page and the link destination web page are the same, the matrix L is a weighted addition of the matrix L _oi and the matrix L _ii It's good if it's a matrix that does. Further, when evaluating the relevance to the search term only for outbound links, the matrix L may be a matrix obtained by weighted addition of the matrix _Loe and the matrix _Loi . , the matrix L may be a matrix obtained by weighted addition of the matrix L _ie and the matrix L _ii . Furthermore, when evaluating the relevance to a search term targeting only external outbound links, the matrix L may be a matrix obtained by weighting the matrix _Loe . When evaluating the property, the matrix L may be a matrix obtained by weighting the matrix L _ie . Similarly, when evaluating the relevance to the search term targeting only the internal outbound links, the matrix L may be a matrix obtained by weighting the matrix L _oi , and targeting only the internal inbound links to the search term. When evaluating the relevance, the matrix L may be a matrix obtained by weighting the matrix L _ii . Also, whether the matrix L is generated by weighted addition of the matrix L _oe , the matrix L _ie , L _oi , or the matrix L _ii , or the matrix L is generated by the matrix L _oe , the matrix L _ie , L _oi , and the matrix L It is also possible to allow the user to specify which weighting of _ii is used for generation each time a web page is evaluated.

(3) In the first embodiment, the evaluation program 26A for causing the CPU 22 of the evaluation device 20A to execute the evaluation method that conspicuously shows the features of the present invention is stored in advance in the hard disk 25 of the evaluation device 20A. However, the evaluation program 26A may be written on a computer-readable recording medium such as a DVD and distributed, or the evaluation program 26A may be distributed by downloading via an electric communication line such as the Internet. This is because by operating a general computer according to the evaluation program 26A distributed in this way, it becomes possible to make the computer function as the evaluation device 20A. Similarly, the evaluation program 26B in the second embodiment may be distributed by writing the evaluation program 26B on a computer-readable recording medium such as a DVD. may be distributed.

(4) In the first embodiment, the designation process, the specification process, the setting process, and the correction process that conspicuously characterize the present invention are realized by software. 20A may be realized. Specifically, designation means for executing the designation process of step SA100 in FIG. 2, specifying means for executing the specifying process of step SA110, setting means for executing the setting process of step SA120, and correction process of step SA130. may be configured by an electronic circuit such as an ASIC, and these electronic circuits may be combined to realize the evaluation device 20A. The same applies to the evaluation device 20B of the second embodiment. That is, each means of specifying means for executing the specifying process of step SB100 in FIG. 7, specifying means for carrying out the specifying process of step SA110 in FIG. The evaluation device 20B may be realized by combining these electronic circuits.

1A, 1B... evaluation system, 10... user terminal, 20A, 20B... evaluation device, 21... communication interface, 22... CPU, 23... RAM, 24... ROM, 25... hard disk, 26A, 26B... evaluation program.

Claims (5)

designating means for designating a web page to be evaluated and a search term likely to be used when searching for the web page to be evaluated;
a web page to be evaluated; a direct link page that is a web page that is directly linked to the web page to be evaluated; and a web that is linked to the web page to be evaluated via one or more other web pages. identifying means for analyzing web page data of each of the indirect link pages, which are pages, and identifying link relationships between the web pages;
setting means for setting, for each web page, an index indicating relevance to the search term for each of the web page to be evaluated, the direct link page, and the indirect link page;
Using at least the index set in the direct link page or indirect link page of the link destination of the internal outbound link or the external outbound link included in the link relationship specified by the specifying means, the web to be evaluated by the setting means correction means for correcting the index set on the page;
An evaluation device comprising: The specifying means is
2. The evaluation apparatus according to claim 1, wherein a hierarchy number N in a link relationship among said web page to be evaluated, said direct link page and said indirect link page is designated. The setting means
3. The index is set for each web page by analyzing web page data of each of the web page to be evaluated, the direct link page, and the indirect link page. Evaluation device. designating means for designating a web page to be evaluated and a search term likely to be used when searching for the web page to be evaluated;
a web page to be evaluated; a direct link page that is a web page that is directly linked to the web page to be evaluated; and a web that is linked to the web page to be evaluated via one or more other web pages. identifying means for analyzing web page data of each of the indirect link pages, which are pages, and identifying link relationships between the web pages;
setting means for setting, for each web page, an index indicating relevance to the search term for each of the web page to be evaluated, the direct link page, and the indirect link page;
a correcting means for correcting the index set for each web page by the setting means according to the link relationship specified by the specifying means;
has
The correcting means is
A diagonal matrix C in which the indices set by the setting means for each web page are arranged in diagonal elements, and a matrix A _n-1 representing a link structure of the n-th layer with the web page to be evaluated as the uppermost stream. and a scalar w _n-1 that is the weight of the link in the n-th layer, a matrix C′ whose elements are the indices after correction is calculated according to formula (10),

There is a relationship of formula (8) between the matrix A _n and the matrix A _n-1 ,

The matrix L in the above formula (8) is
A matrix L _oe representing a link in the direction from the link source web page to the link destination web page, where the domains of the link source web page and the link destination web page are different, and the transposed matrix L _ie of the matrix L _oe , the domain of the link source web page and the link destination web page are the same, the matrix L _oi representing the link in the direction from the link source web page to the link destination web page, and the transposed matrix L of the matrix L _oi A matrix obtained by weighted addition of a plurality of _ii , or a matrix obtained by weighting any of the matrix L _oe , the matrix L _ie , the matrix L _oi , and the matrix L _ii ,
An evaluation device characterized by: a designating step in which designating means designates a web page to be evaluated and a search term likely to be used when searching for the web page to be evaluated;
The identifying means includes the web page to be evaluated, a direct link page that is a web page directly linked to the web page to be evaluated, and one or more other web pages to the web page to be evaluated. an identifying step of analyzing the web page data of each of the indirect link pages, which are web pages to be linked, and identifying the link relationship between the web pages;
a setting step in which setting means sets, for each of the web page to be evaluated, the direct link page, and the indirect link page, an index indicating the relevance between the web page and the search term for each web page;
The correction means uses at least the index set to the direct link page or the indirect link page of the link destination of the internal outbound link or the external outbound link included in the link relationship specified in the specifying step, by the setting means a correction step of correcting the index set for the web page to be evaluated;
An evaluation method characterized by comprising

Images