Proliferation and Detection of Blog Spam

摘要

The ease of posting comments and links in blogs has attracted spammers as an alternative venue to conventional email. An experimental study investigates the nature and prevalence of blog spam. Using Defensio logs, the authors collected and analyzed more than one million blog comments during the last two weeks of June 2009. They used a support vector machine (SVM) classifier combined with heuristics to identify spam posters' IP addresses, autonomous system numbers (ASN), and IP blocks. Experimental results show that more than 75 percent of blog comments during the reporting period are spam. In addition, the results show that blog spammers likely operate from a few colocation facilities.