Discriminating Biased Web Manipulations in Terms of Link Oriented Measures

摘要

In this paper, we present a link oriented measuring method to discriminate the manipulated web pages effectively. We define the label of an edge as having a link context and a similarity measure between link context and target page. By suggesting an assessing measure based on singular value decomposition, it is explained that our proposed method can effectively detect the manipulated web pages. We, however, extend the SVD as an assessment measure to detect the rank-manipulated pages. In the experiment, the LOD method reduced about 17% amount of the rank that is minimum 209.4% higher than not manipulated web pages. Using this proposed approach, the chance of manipulated web pages getting high ranks than deserved can be discriminated effectively.