Using Spam Farm to Boost PageRank

Today people have become more and more dependent on search engines such as Google, Yahoo, and MSN, etc., for their information needs. Web spamming has emerged to take the economic advantage of high search rankings and threatened the accuracy and fairness of those rankings. Understanding spamming techniques is essential for evaluating the strength and weakness of a ranking algorithm, and for fighting against web spamming. In this paper, we identify the optimal spam farm structure under some realistic assumptions in the single target spam farm model. Our result contradicts the optimal spam farm claimed in [7], the proof of which is fundamentally flawed. We also characterize the optimal spam farms under some additional and realistic constraints, which the spammer may deploy to disguise the spam farm by deviating from the unconstrained optimal structure. In the simulation, we show that the optimal spam farm can significantly boost the PageRank score of a target page. In particular, the boosting effect is more significant for target pages with low PageRank scores. Furthermore, by using web pages with higher PageRank scores as boosting pages, the spammer can get better boosting effect too.