Methods and systems for identifying manipulated articles

• US 7,302,645 B1
• Filed: 12/10/2003
• Issued: 11/27/2007
• Est. Priority Date: 12/10/2003
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Henzinger, Monika, Franz, Alexander Mark
• Primary Examiner(s): Nguyen; Cao (Kevin)
• Application Number:
  US10/732,048
• Time in Patent Office: 1,448 Days
• Field of Search: 715/738, 715/745, 715746-748, 715/751, 715/805, 715/854
• US Class Current: 715/738
• CPC Class Codes: G06F 16/355 Class or cluster creation or modification
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques

Abstract

Systems and methods that identify manipulated articles are described. In one embodiment, a search engine implements a method comprising determining at least one cluster comprising a plurality of articles, analyzing signals to determine an overall signal for the cluster, and determining if the articles are manipulated articles based at least in part on the overall signal.

Claims

1. A method, comprising:

determining at least one cluster comprising a plurality of articles;

analyzing signals associated with one or more articles in the plurality of articles to determine an overall signal for the cluster; and

determining if articles in the plurality of articles are manipulated articles based at least in part on the overall signal;

wherein determining the at least one cluster comprises computing a dense bipartite subgraph of articles comprising doorway articles and target articles, wherein the doorway articles contain links to the target articles.

2. The method of claim 1, wherein determining the at least one cluster comprises identifying all of the documents on a host.

3. The method of claim 2, wherein the host is likely to contain manipulated articles.

4. The method of claim 1, wherein the overall signal represents a likelihood that the articles are manipulated articles.

5. The method of claim 1, wherein the overall signal is compared to a threshold to determine whether the articles are manipulated articles.

6. The method of claim 1, wherein the overall signal is used at least in part in ranking an article in response to a search query.

7. The method of claim 1, wherein the signal comprises at least one of outside signals and document signals.

8. A computer-implemented method comprising:

forming a cluster of documents from a plurality of network-accessible documents by identifying a dense bipartite subgraph from the plurality of network-accessible documents, the dense bipartite subgraph comprising a first set of doorway documents and a second set of target documents, wherein doorway documents in the first set have links to target documents in the second set;

analyzing a plurality of documents in the cluster of documents to determine an overall value for the cluster; and

when the overall value is greater than a threshold value, marking at least one of the documents in the cluster as a manipulated article.

9. The computer-implemented method of claim 8, wherein marking the at least one document as a manipulated article comprises identifying the at least one document as a document that has been manipulated to increase the document'"'"'s ranking by a search engine.

10. The computer-implemented method of claim 8, wherein marking the at least one document comprises preventing the at least one document from being included in a search result set provided by a search engine in response to a search query.

11. The computer-implemented method of claim 8, further comprising marking each document in the cluster as a manipulated article.

12. The computer-implemented method of claim 8, wherein forming the cluster of documents further comprises including documents in the cluster that are stored on a host that stores other documents included in the dense bipartite subgraph.

13. The computer-implemented method of claim 8, further comprising marking as manipulated articles other documents that are stored on a host that stores the at least one document in the cluster.

14. The computer-implemented method of claim 8, wherein analyzing a plurality of documents in the cluster of documents comprises determining if a target document in the cluster of documents has more than a threshold number of links to guestbook documents that are included in the doorway documents.

15. The computer-implemented method of claim 8, wherein analyzing a plurality of documents in the cluster of documents comprises determining if more than a threshold number of the plurality of documents in the cluster of documents each has an individual document signal that indicates that the document is likely to be a manipulated article.

16. The computer-implemented method of claim 15, wherein the individual document signal is an indication that the corresponding document has a large number of keywords without a proportional number of sentences.

17. The computer-implemented method of claim 15, wherein the individual document signal is an indication that the corresponding document includes meta tags having a large number of repeated keywords.

18. The computer-implemented method of claim 15, wherein the individual document signal is an indication that the corresponding document includes a script configured to redirect a user upon access of the corresponding document.

19. The computer-implemented method of claim 15, wherein the individual document signal is an indication that the corresponding document includes more than a threshold amount of text having the same color as a background color of the corresponding document.

20. The computer-implemented method of claim 15, wherein the individual document signal is an indication that the corresponding document includes more than a threshold number of unrelated links.

21. The computer-implemented method of claim 15, wherein the individual document signal is an indication that a link structure of the corresponding document has recently changed or that ownership of a web site with which the corresponding document is associated has recently changed.

22. The computer-implemented method of claim 15, wherein the individual document signal is an indication that the corresponding document includes more than a threshold number of links that lack corresponding anchor text.