Investigating Spam Mass Variations for Detecting Web Spam

Abstract

In this paper, we investigate variations of Spam Mass for filtering web spam. Firstly, we propose two strategies for designing new variations of the Spam Mass algorithm. Then, we perform experiments among different versions of Spam Mass using WEBSPAM-UK2006 data set. Finally, we show improvement through proposed strategy by up to 1.33 times in recall and 1.02 times in precision over the original version of Spam Mass.