JP2007183825A - Score derivation system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To enable score tabulation by author, in a score derivation system for ranking the popularity stakes of Internet sites or the like, while preventing deterioration of accuracy by spam link, etc. with a reduced load to a server by shortening processing time. <P>SOLUTION: This system comprises a means for extracting, from a site on a network, at least identification information of a page, identification information of a page linked from this page, and information for identifying a creation subject of this page; a means for grouping the extracted information to creation subject-segregated link relations based on the information for identifying the creation subject; and a means for calculating a score based on the grouped creation subject-segregated link relations. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a score derivation system for ranking the popularity etc. of Internet sites.

  Recently, blogs on the Internet (Blog, Weblog: a diary Web site operated by individuals and groups of people and updated daily) have attracted attention, and there are many blogs. The blog is composed of a plurality of HTML (Hyper Text Markup Language) pages.

  A service for ranking and introducing popular blogs is also available, and a score derivation method for ranking has been proposed (see, for example, Patent Documents 1 and 2).

  FIG. 1 is a diagram showing an outline of conventional score derivation called “Page Rank” disclosed in Patent Documents 1 and 2. In FIG.

In FIG. 1, a link is made from page A to page B and page C as indicated by a line with an arrow, a link is made from page B to page C, and a link is made from page C to page A. Then, the score (rank) of each page in the (i + 1) th calculation is
r _{i + 1} (A) = r _i (C)
r _{i + 1} (B) = r _i (A) / 2
r _{i + 1} (C) = r _i (A) / 2 + r _i (B)
become that way. This is based on the premise that the score of a certain page depends on the score of the page linked to that page. When there are a plurality of links from one page, the score added to the linked page is divided by the number of links.

For example, each score of the first time is r ₁ (A) = 1
r ₁ (B) = 1
r ₁ (C) = 1
, Each score for the second time is
r ₂ (A) = 1
r ₂ (B) = 0.5
r ₂ (C) = 1.5
And each score for the third time is
r ₃ (A) = 1.5
r ₃ (B) = 0.5
r ₃ (C) = 1
By repeating (looping) such a calculation for a predetermined number of times (in this example, several tens of times),
r _∞ (A) = 1.2
r _∞ (B) = 0.6
r _∞ (C) = 1.2
Will converge to.
US Pat. No. 6,285,999 US Pat. No. 6,799,176

  Although the example shown in FIG. 1 is based on three simplified pages, it is necessary to perform a considerable amount of calculation processing to obtain a convergence value. In fact, a large number of pages are processed. As a result, there is a problem that processing time is required and the load on the server is heavy. In addition, in order to take into account random links to unexpected pages, a certain percentage of each page's score (about 10%) is evenly added to the score of other pages. Furthermore, the burden of calculation processing increases.

  On the other hand, intentional links (spam links, spam trackbacks, etc.) are also provided between pages of my blog in an attempt to increase the ranking of my blog. Score up could not be prevented.

  Furthermore, although it is desired to rank popularity etc. for each author of the blog, since the conventional score derivation can be performed only in units of pages, it is necessary to further add up the scores obtained in units of pages for each author.

  The present invention has been proposed in view of the above-mentioned conventional problems, and its object is to reduce processing time and reduce the burden on the server, and to prevent deterioration in accuracy due to spam links and the like. It is another object of the present invention to provide a score derivation system capable of collecting scores for each author.

  In order to solve the above-mentioned problem, in the present invention, as described in claim 1, from a site on a network, at least page identification information, page identification information linked from the page, and Means for extracting information for identifying the author of the page, means for grouping from the extracted information into a link relationship for each author based on the information for identifying the author, and grouped The gist of the present invention is a score derivation system comprising means for calculating a score based on the link relationship for each of the above-mentioned creation subjects.

  Further, as described in claim 2, in the score derivation system according to claim 1, the information for identifying the creation subject may be an ID included in the RSS.

  In addition, as described in claim 3, in the score derivation system according to claim 1, the information for identifying the creator may be a member ID of the creator on the network service. .

  Further, as described in claim 4, in the score derivation system according to any one of claims 1 to 3, the means for grouping into the link relation for each creation subject is the extracted above Means for extracting correspondence information between a page and the page to which the page is linked from the information, means for extracting correspondence information between the page and information for identifying a creation entity to which the page belongs, from the extracted information, Means for extracting link correspondence information for each creation subject from correspondence information between the page and the page to which the page is linked and correspondence information between the page and information identifying the creation subject to which the page belongs can be provided.

  Further, as described in claims 5 and 6, it can be configured as a score derivation method.

  In the score derivation system of the present invention, the processing time can be shortened and the burden on the server can be reduced, the accuracy can be prevented from decreasing due to spam links and the like, and the score can be tabulated for each author.

  Hereinafter, preferred embodiments of the present invention will be described.

  FIG. 2 is a diagram illustrating a configuration example of the score derivation system 100 according to the embodiment of the present invention. In FIG. 2, the score derivation system 100 includes at least identification information of a page from content of a site to be measured on the Internet 200, identification information of a page linked from the page, and information for identifying a creation subject of the page. A content extraction unit 110 that extracts and stores the data in the data storage unit 120, and a grouping processing unit 130 that obtains information from the data storage unit 120 and groups into a link relationship for each creation subject based on information that identifies the creation subject. And a score calculation processing unit 140 that calculates a score based on the link relationship for each grouped creator. As page identification information, a URL (Uniform Resource Locator) can be used. Further, as information for identifying the creator, an ID (RSS-ID) included in RSS (Rich Site Summary, Really Simple Syndication, RDF (Resource Description Framework) Site Summary) that provides blog summary information is used. it can. The RSS-ID can be acquired from the link information to the RSS embedded in the source information. Further, the member ID of the creator on the network service can be used as information for identifying the creator.

  FIG. 3 is a diagram illustrating an example of information stored in the data storage unit 120. The URL of the page, the URL of the page to which the page is linked, and the RSS-ID of the page (link source) are included in the number of pages. It is. Note that the content extraction unit 110 excludes information on a page that is not a blog (a page that does not include a tag indicating link information to the RSS in the source) when extracting page information. For example, even if a link is made from page A to page C, if page C is not a blog, information about page C is not extracted as information on the page itself or information on the link destination. By doing so, subsequent processing can be lightened.

  FIG. 4 is a diagram illustrating a detailed configuration example of the grouping processing unit 130, and includes a link extraction unit 131, an RSS extraction unit 132, and an inter-RSS link extraction unit 133.

  In FIG. 4, a link extraction unit 131 extracts correspondence information between a page and a page to which the page is linked from information in the data storage unit 120, and generates a page / link page correspondence table T1. In the example shown in the figure, page A is linked to pages B and D, and page B is linked to pages E, F, and G. Although the link from page A to page C is made, page C is assumed not to be a blog, so information on page C does not come out.

  The RSS extraction unit 132 also extracts correspondence information between a page and an RSS-ID to which the page belongs from information in the data storage unit 120, and generates a page / RSS correspondence table T2 or an RSS / page correspondence table T3. In the illustrated example, pages A and D belong to RSS1, and pages B, E, and F belong to RSS2. Also here, since the page C linked from the page A is assumed not to be a blog, the information of the page C does not come out.

  Next, the RSS link extracting unit 133 extracts link correspondence information for each RSS-ID from the page / link page correspondence table T1 and the page / RSS correspondence table T2 or the RSS / page correspondence table T3, and the RSS / link RSS. A correspondence table T4 is generated. In the example of illustration, it has shown that it links from RSS1 to RSS2, and is linked from RSS2 to RSS1, RSS3.

  And the score calculation process part 140 replaces the page in the conventional score derivation method shown in FIG. 1 with the group for every RSS-ID based on RSS / link RSS corresponding | compatible table T4, and calculates a score.

  FIG. 5 is a diagram showing an example of links for which scores are calculated in the present invention. Assuming that there is a blog of author 1, a blog of author 2, and a blog of author 3, a plurality of links are provided between pages in each blog. Even if it exists, by grouping miscellaneous pages into blogs for each author, only the links L1, L2, and L3 between the blogs are subject to score calculation, and the links in the blog are ignored. Also, links to pages other than blogs are ignored.

  Therefore, it is possible to shorten the processing time for calculating the score and lighten the burden on the server, and to prevent a decrease in accuracy due to a spam link or the like. Moreover, the score for every creator can be calculated | required directly.

FIG. 6 is a diagram showing a difference in processing time between the conventional technique and the present invention. In FIG. 6, the loop processing time _TN for the page is conventionally required, but in the present invention, the grouping processing time T _m and the loop processing time T _M for the group are required. In the following, the magnitude relationship will be considered using mathematical formulas.

Each constant is defined as follows.
n: number of pages (nodes) m: number of groups C: grouping process time per group (constant)
N: Time for one loop process per page or group (constant)
β: Loop processing repetition rate α: Processing time reduction rate by grouping (0 <α ≦ 1)
Using these, the time _{TN of the} loop processing for the conventional page is expressed as follows:
T _N = β · n · N
It becomes.

The grouping processing time T _m in the present invention is:
T _m = m · C
Next, time T _M of the loop processing for the group,
T _M = β ・ m ・ α ・ N
It becomes.

Here, when the processing time of the present invention is subtracted from the conventional processing time,
T _N − (T _m + T _M ) = β · n · N−m · C−β · m · α · N
= Β · N (n−m · α) −m · C
It becomes.

Experience has shown that a blog has more than 10 pages,
n ≧ 10m
N = 10m, the lowest value
After all,
T _N − (T _m + T _M ) = β · N (10 m−m · α) −m · C
= Β · N · m (10−α) −m · C
It becomes.

Furthermore, consider the worst state (state where there is no reduction in processing time) as α,
α = 1
Then,
T _N − (T _m + T _M ) = β · N · m · 9−m · C
= M (9 · β · N-C)
It becomes.

Here, since N and C are values of approximately the same order, 9 · β · N is a very large value compared to C,
T _N − (T _m + T _M ) >> 0
However, it can be seen that there is a significant time saving effect.

  The present invention has been described above by the preferred embodiments of the present invention. While the invention has been described with reference to specific embodiments, various modifications and changes may be made to the embodiments without departing from the broad spirit and scope of the invention as defined in the claims. Obviously you can. In other words, the present invention should not be construed as being limited by the details of the specific examples and the accompanying drawings.

It is a figure which shows the outline | summary of the conventional score derivation. It is a figure showing an example of composition of a score derivation system concerning one embodiment of the present invention. It is a figure which shows the example of the storage information of a data storage part. It is a figure which shows the detailed structural example of a grouping process part. It is a figure which shows the example of the link used as the object of score calculation in this invention. It is a figure which shows the difference in the processing time of the past and this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Score derivation system 110 Content extraction part 120 Data storage part 130 Grouping process part 131 Link extraction part 132 RSS extraction part 133 Link extraction part between RSS 140 Score calculation processing part 200 Internet T1 page / link page correspondence table T2 page / RSS correspondence Table T3 RSS / page correspondence table T4 RSS / link RSS correspondence table

Claims (6)

Means for extracting at least page identification information, page identification information linked from the page, and information for identifying the creator of the page from a site on the network;
Means for grouping into a link relationship for each creator based on the information identifying the creator from the extracted information;
A score derivation system comprising: means for calculating a score on the basis of the link relation for each of the grouped creators. The score derivation system according to claim 1,
The score derivation system, wherein the information for identifying the creation subject is an ID included in RSS. The score derivation system according to claim 1,
The score derivation system, wherein the information for identifying the creator is a member ID of the creator on the network service. In the score derivation system according to any one of claims 1 to 3,
The means for grouping into the link relationship for each creator is
Means for extracting correspondence information between a page and a page linked to the page from the extracted information;
Means for extracting correspondence information between a page and information for identifying a creation subject to which the page belongs, from the extracted information;
Means for extracting link correspondence information for each creation subject from correspondence information between the page and the page to which the page is linked and correspondence information between the page and information for identifying the creation subject to which the page belongs. A characteristic score derivation system. Extracting at least page identification information, page identification information linked from the page, and information identifying the creation subject of the page from a site on the network;
Grouping from the extracted information into a link relationship for each creator based on the information identifying the creator;
And a step of calculating a score on the basis of the link relation for each of the grouped creators. The score derivation method according to claim 5,
The process of grouping into the link relationship for each creator is
Extracting the correspondence information between the page and the page linked to the page from the extracted information,
Extracting correspondence information between a page and information for identifying a creation subject to which the page belongs, from the extracted information,
A step of extracting link correspondence information for each creation subject from correspondence information between the page and the page linked to the page and correspondence information between the page and information identifying the creation subject to which the page belongs. A characteristic score derivation method.

Images