Method for detecting search engine cheat based on small sample set

Abstract

The invention relates to Internet information retrieval and discloses a method for detecting Internet cheats on the basis of a small sample set to strike search engine cheating behaviors which become severe increasingly. The invention uses the iterative execution of self-learning on the basis of a classifier and the linking learning process on the basis of an Internet topological structure to continuously expand a training set aiming at the problem that a detection sample has high collecting cost to realize that search engine cheats are detected under the small sample set, an integrated down sampling strategy is adopted in the identification process, and information which is contained in websites which exist widely on Internet and have high reputation is fully used. Finally, label transmission along the Internet topological structure and on the basis of prediction cheating degree is carried out to realize the optimization of detection results. Experiments are used to show that the method can effectively detect the cheating behaviors.

Description

technical field [0001] The invention relates to the technical fields of information retrieval and search engines, and relates to a method for detecting cheating of search engines under small samples. Background technique [0002] As the largest repository of information ever created, the Internet is still growing exponentially. Internet search has become a part of people's daily life. According to a report released by CNNIC in July 2006, search engines accounted for 66.3% of the Internet users' most frequently used Internet services. [0003] Scholars such as N. Eiron used the famous PageRank algorithm to sort 100 million web pages, and found that 11 of the top 20 websites were pornographic websites, and these websites obtained high rankings by tampering with hyperlinks. According to the survey of the US Business Research Bureau, in 2006, the e-commerce sales in the United States reached 114.1 billion US dollars, an increase of 22.7% compared with 93 billion US dollars in 2...

AI Technical Summary

Problems solved by technology

[0008] In order to solve the problem that the acquisition cost of samples required by existing machine learning methods is high, and traditional learning algorithms are difficult to achieve good results in unbalanced sample learning, the purpose of this invention is to reduce the human resources required for obtaining samples and reduce costs Obtain good results in unbalanced sample learning, for this reason the present invention provides a kind of search engine Web cheating detection method based on small sample set

Images