Abstract:

Spam pages seek to boost their ranking positions and thus earn profit mainly through spam links. Based on the analytical features of spam links, we introduce link similarity and spam rate as two indexes for spam page judgment. Inspired by the Bad Rank algorithm, we calculate link similarity and Spam rate by iteration from the seed set of spam pages, set the weights in accordance with the relationship of link pointing from the seed set of spam pages, and measure the pages to be judged. AntiTrust Rank and other relevant approaches are adopted to make comparison with the traditional comparative methods. Experimental results prove the advantage of our approach over the traditional methods.

Key words: spam page;spam link;link similarity;spam rate;weight coefficient