From Good to Bad Ones: Making Spam Detection Easier

Previous researches of the anti-spamming have proved that results can be greatly improved when bad seeds are used together with good ones. However, how to select bad seeds efficiently is a big challenge. In this paper we discuss how to select bad seeds based on good seeds selection. The experiments running on over 13M Web pages show that our method is practical and time-saving. Moreover, the selected bad seeds can enhance the performance of a good seed set on effectively filtering out spam from normal pages.