JP2010066980A - Spam blog detection device, spam blog detection method, and program - Google Patents

Abstract

A spam blog detection apparatus, a spam blog detection method, and a program capable of accurately detecting a spam blog are provided.
A spam blog detection apparatus 1 includes a link destination evaluation unit that evaluates a website linked from a blog, a blog evaluation unit that evaluates a blog using a website evaluation result, and a blog evaluation. And determining means for determining whether or not the blog is a spam blog.
[Selection] Figure 2

Description

  The present invention relates to a spam blog detection device, a spam blog detection method, and a program.

Currently, an advertising method called an affiliate program has appeared on the Internet. The affiliate program is a method in which an advertisement medium owner places an advertiser's advertisement on his / her own web page, blog, etc., and clicks on the advertisement to connect to the advertiser's site. A blog is a website such as a diary published on the Internet.
By the way, among blogs that post information on the Internet, there are many spam blogs (splog: Spam Weblog) arbitrarily created by advertisers. In spam blogs, articles are usually created by embedding only meaningless keywords in large quantities, or by copying sentences from other blogs or news articles without permission. For this reason, many spam blogs are meaningless as content.

Non-Patent Document 1 proposes the following three methods as a spam blog determination method.
(1) A technique for determining a spam blog as a link to an affiliate program.
(2) A method of calculating a page rank from the number of links from other websites and determining that a site with a high page rank is not a spam blog.
(3) A technique for determining a blog in which words frequently appear in a spam blog frequently appear as a spam blog.
Yu-Ru Lin, Sundaram, H., Yun Chi, Tatemura, J., Tseng, B., “Splog Detection using Content, Time and Link Structures”, Multimedia and Expo, 2007 IEEE International Conference on

However, since there is a case where a general blog is linked to an affiliate program, there is a problem that there is a possibility of erroneous detection only by looking at the presence or absence of a link as in the method (1) described in Non-Patent Document 1. is there.
In addition, some spam blogs increase the page rank by creating a number of sites that link to blogs created by themselves, so that the technique (2) described in Non-Patent Document 1 is used. The number of backlinks doesn't mean that the value of a site is determined.
In the method (3) described in Non-Patent Document 1, detection is avoided if a word that frequently appears in a spam blog is removed. Moreover, since it is thought that those words appear also in a general site, we are anxious about false detection.
The present invention has been made in view of the above points, and an object of the present invention is to provide a spam blog detection device, a spam blog detection method, and a program capable of detecting a spam blog with high accuracy.

  The present invention has been made to solve the above problems, and one aspect of the present invention uses link destination evaluation means for evaluating a website linked from a blog and the evaluation result of the website. A spam blog detection apparatus comprising: a blog evaluation unit that evaluates the blog; and a determination unit that determines whether the blog is a spam blog based on the blog evaluation.

Further, according to one aspect of the present invention, in the spam blog detection apparatus, blog information acquisition means for acquiring information on a plurality of blogs, and information on websites linked from the acquired blog information from the acquired blog information Link destination information acquisition means for acquiring the information, wherein the link destination evaluation means evaluates the website from information of a blog linked to the website.

  Further, according to an aspect of the present invention, in the spam blog detection apparatus, the link destination information acquisition unit calculates the number of blogs linked to the affiliate program providing site for each identification information issued by the affiliate program. The link destination evaluation means evaluates the providing site using the number of blogs for each identification information.

  One embodiment of the present invention is characterized in that, in the spam blog detection apparatus, a website linked from the blog is re-evaluated using the evaluation of the blog.

  According to another aspect of the present invention, there is provided a step of evaluating a website linked from a blog, a step of evaluating the blog using the evaluation result of the website, and the blog based on the evaluation of the blog. And a step of determining whether or not the message is a spam blog.

  According to another aspect of the present invention, there is provided a step of evaluating a website linked from a blog, a step of evaluating the blog using the evaluation result of the website, and the blog based on the evaluation of the blog. Is a program for causing a computer to execute a step of determining whether or not a spam blog.

  According to the present invention, since the blog is evaluated using the evaluation result of the website linked from the blog, it is possible to accurately detect the spam blog corresponding to the change of the blog.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a conceptual diagram showing a concept of processing of a spam blog detection apparatus according to an embodiment of the present invention.
Blogs can be created and edited freely by creators, so ordinary people can use affiliate programs and link to arbitrary sites. In addition, the content of the blog is assumed to change frequently. Therefore, the spam blog detection apparatus of the present invention evaluates affiliate programs and advertisement providing sites with relatively little change in content, and detects spam blogs using the evaluation results for blog evaluation.
The spam blog detection apparatus first acquires blog information of a plurality of blogs. The blog information includes a blog URL (Uniform Resource Locator), a link destination URL from the blog, the number of links from the blog, and the like. Next, the spam blog detection device evaluates the affiliate program providing site or the advertisement providing site linked from the blog. And a spam blog detection apparatus evaluates a blog using the evaluation result, and detects a spam blog.

FIG. 2 is a block diagram showing a configuration of the spam blog detection apparatus 1 in the present embodiment.
The spam blog detection apparatus 1 includes a blog information acquisition unit 11, a link destination information acquisition unit 12, an information analysis unit 13, and a determination unit 14.
The blog information acquisition unit 11 acquires blog information of a plurality of blogs using information on blog articles such as RSS (Rich Site Summary) distributed by an application service provider and new blog information.

The link destination information acquisition unit 12 acquires information on a linked site (link destination) from the acquired blog information. Specifically, when the link destination is an affiliate program providing site, the link destination information acquiring unit 12 uses the ID to link to the providing site for each ID (identification information) issued by the providing site. Calculate the number of active blogs (link source blogs). The ID is included in the blog information acquired by the blog information acquisition unit 11. A case where the blog information acquisition unit 11 acquires the blog information of the blogs A to I illustrated in FIG. 1 will be described as an example. The blogs referring to the affiliate program providing sites are A, B, C, D, E, and H. When the IDs of blogs A, B, and C are all “a”, the ID of blog D is “d”, and the IDs of blogs E and H are “e”, the number of link source blogs with ID “a” is 3 become. Also, the number of link source blogs with ID “d” is 1. Also, the number of link source blogs with ID “e” is two.
The same processing is performed when the link destination is an advertisement providing site.

The information analysis unit 13 performs blog evaluation, that is, scoring (scoring) based on the information acquired by the blog information acquisition unit 11 and the link destination information acquisition unit 12. The higher the score (score), the more likely it is a spam blog.
The determination unit 14 determines whether the blog related to the blog information acquired by the blog information acquisition unit 11 is a spam blog based on the evaluation result of the information analysis unit 13. Specifically, the determination unit 14 determines a blog having a score higher than a predetermined threshold set in advance as a spam blog.

Next, the operation of the above-described spam blog detection apparatus 1 according to the present embodiment will be described. FIG. 3 is a flowchart showing an operation procedure of the spam blog detection apparatus 1 according to the present embodiment.
First, in step S1, the blog information acquisition unit 11 acquires blog information of a plurality of blogs. At this time, the blog information acquisition unit 11 extracts a plurality of blogs from RSS or new blog information provided by the application service provider, and acquires the blog information of the extracted blogs.
Next, in step S2, the link destination information acquisition unit 12 acquires the link destination information of the blog extracted in step S1.
Next, in step S3, the information analysis unit 13 scores each blog based on the information acquired in steps S1 and S2. Details of the processing of the information analysis unit 13 will be described later.
Next, in step S4, the determination unit 14 determines whether each blog is a spam blog. Specifically, the determination unit 14 determines that a blog whose score is higher than a predetermined threshold is a spam blog.

FIG. 4 is a flowchart showing an operation procedure of the information analysis unit 13 in the present embodiment.
First, in step S11, the information analysis unit 13 evaluates each blog extracted in step S1 described above. The score score _{bpre of} each blog is calculated by the following equation (1), for example. Here, α _i is a feature amount, and w1 _i is a weight of each feature amount. The weight w1 _i is a preset value. N is the number of feature quantities. The feature amount includes the number of links from the blog (α ₁ ), the type of link destination (α ₂ ), the posting time of the article (α ₃ ), the posting frequency (α ₄ ), and the like. Expression (1) is an expression obtained by adding the scores of the feature amounts.

Here, the score increases as the number of links increases. The type of link destination is a value obtained by adding all the scores of sites linked from the blog. Here, a score is set in advance for each site. For example, the score of a generally famous site is set low, and the score of an affiliate site is set high. If a site that has not been set is linked, the score is set to a standard or slightly higher than the standard.
The score of the posting time of the article is determined based on the distribution of the posting time. For example, if the posting is made at the same time every day, the score is increased. This is because it is assumed that articles are automatically created by the program. Also, the score is increased when the posting time variation is very large. This is because it is assumed that there is a high possibility of being a spam blog when the life time zone cannot be specified. Also, the lower the posting frequency, the higher the score.

Next, in step S12, the information analysis unit 13 evaluates all affiliate programs and advertisement providing sites linked from each blog. The score score _a of each affiliate program is calculated by the following equation (2), for example. Here, β _i is a feature amount, and w2 _i is a weight of each feature amount. L is the number of feature values. Also, score _{bpre, i} is the score of the blog using the affiliate program, and m is the number of blogs using the affiliate program. Also, w _score is a weight for the blog score. Expression (2) is an expression in which each feature amount and the blog score are added.

The feature amount includes the number of link source blogs (β ₁ ) for each ID issued by the affiliate program. For example, when links are made from many blogs with one ID, the score is high. Further, the service (β ₂ ) provided by the affiliate program may be used as the feature amount. For example, a different score is set for each sale product (for example, book, clothing, food). The evaluation of the advertisement providing site is performed in the same way.

Next, in step S13, the information analysis unit 13 reevaluates each blog from the affiliate program calculated in step S12 and the score of the advertisement providing site. The score score _{b of} each blog is calculated by the following equation (3), for example. Here, score _{a, i} is the score of the affiliate program or advertisement providing site linked from the blog, and w score _{, i} is the weight of each affiliate program or advertisement providing site. K is the number of affiliate programs and advertisement providing sites linked from the blog. Expression (3) is an expression obtained by adding each feature amount and the score of the affiliate program and the advertisement providing site.

Here, the weight w _{score, i} of each affiliate program and the advertisement providing site is set in advance. For example, the weight w _{score, i} of a generally famous affiliate program is set relatively light. Note that the weight w _{score, i} of an affiliate program for which no weight is set is set to a predetermined value set in advance.

Next, in step S14, the information analysis unit 13 feeds back the blog score calculated in step S13 to the evaluation of the affiliate program and the advertisement providing site. That is, the affiliate program and the advertisement providing site are reevaluated using the score calculated in step S13. The score _{scoreback of the} affiliate program is calculated by the following equation (4), for example. _However, score _{b, i} is the score of the blog you are using an affiliate _{program, w scoreb} is a weight for the score of the blog. Expression (4) is an expression in which each feature amount and the blog score are added. The re-evaluation of the advertisement providing site is performed in the same way.

  In step S15, the information analysis unit 13 reevaluates each blog using the above-described equation (3) using the affiliate program calculated in step S14 and the score of the advertisement providing site, and ends the process.

  As described above, according to the present embodiment, the evaluation of the affiliate program with relatively little change in content is used for the evaluation of the blog. Therefore, it is possible to perform spam blog determination with higher accuracy corresponding to the blog situation change.

Further, by recording a program for realizing the function of the spam blog detection device shown in FIG. 2 on a computer-readable recording medium, causing the computer system to read and execute the program recorded on the recording medium, Spam blog detection processing may be performed. Here, the “computer system” may include 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 “computer-readable recording medium” means a flexible disk, a magneto-optical disk, a ROM, a writable nonvolatile memory such as a flash memory, a portable medium such as a CD-ROM, a hard disk built in a computer system, etc. This is a storage device.

Further, the “computer-readable recording medium” means a volatile memory (for example, DRAM (Dynamic DRAM) in a computer system that becomes a server or a client when a program is transmitted through a network such as the Internet or a communication line such as a telephone line. Random Access Memory)), etc., which hold programs for a certain period of time.
The program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line.
The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, and what is called a difference file (difference program) may be sufficient.

  As described above, the embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to the above, and various design changes and the like can be made without departing from the scope of the present invention. It is possible to

It is the conceptual diagram which showed the concept of the process of the spam blog detection apparatus by one Embodiment of this invention. It is a block diagram which shows the structure of the spam blog detection apparatus in this embodiment. It is a flowchart which shows the procedure of the operation | movement of the spam blog detection apparatus in this embodiment. It is a flowchart which shows the procedure of operation | movement of the information analysis part in this embodiment.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Spam blog detection apparatus 11 ... Blog information acquisition part 12 ... Link destination information acquisition part 13 ... Information analysis part 14 ... Determination part

Claims (6)

Link destination evaluation means for evaluating websites linked from blogs,
A blog evaluation means for evaluating the blog using the evaluation result of the website;
Determining means for determining whether the blog is a spam blog based on the evaluation of the blog;
A spam blog detection device comprising: Blog information acquisition means for acquiring information of a plurality of blogs;
From the acquired blog information, link destination information acquisition means for acquiring website information linked from the blog,
With
The spam blog detection apparatus according to claim 1, wherein the link destination evaluation unit evaluates the website from information of a blog linked to the website. The link destination information acquisition means calculates the number of blogs linked to the affiliate program providing site for each identification information issued by the affiliate program,
The spam blog detection device according to claim 2, wherein the link destination evaluation unit evaluates the provided site using the number of blogs for each identification information.   The spam blog detection device according to any one of claims 1 to 3, wherein a website linked from the blog is re-evaluated using the evaluation of the blog. Evaluating websites linked from your blog;
Performing an evaluation of the blog using an evaluation result of the website;
Determining whether the blog is a spam blog based on the evaluation of the blog;
A spam blog detection method characterized by comprising: Evaluating websites linked from your blog;
Performing an evaluation of the blog using an evaluation result of the website;
Determining whether the blog is a spam blog based on the evaluation of the blog;
A program that causes a computer to execute.

Images