After the seminal paper of Kleinberg [1] and the introduction of PageRank [2], there has been a surge of research activity in the area of web mining using link analysis algorithms. Subsequent to the first generation of algorithms, a significant amount of improvements and variations appeared. However, the issue of stability has received little attention in spite of its practical and theoretical implications. For instance, the issue of "link spamming" is closely related to stability: is it possible to boost up the rank of a page by adding/removing few nodes to/from it? In this paper, we study the stability aspect of various link analysis algorithms concluding that some algorithms are more robust than others. Also, we show that those unstable algorithms may become stable when they are properly "randomized".
展开▼
