Securing Reputation Systems in the Cyber Space
Word-of-mouth, one of the most ancient mechanisms in the history of human society, is gaining new significance in the Internet. The online reputation systems, also known as the on-line feedback mechanisms, are creating large scale, virtual word-of-mouth networks in which individuals share opinions and experiences on a wide range of topics, including products, companies, digital content and even other people. Reputation systems are having increasing influence on purchasing decision of consumers and online digital content distribution. Meanwhile, the manipulation of such systems is rapidly growing. Dealing with unfair ratings in online feedback-based rating systems has been recognized as an important and challenging problem.
Different from all previous approaches, we develop an anomaly detection scheme, named TAUCA, for feedback-based reputation systems. Here, TAUCA is the abbreviation for joint Temporal And User Correlation Analysis. In particular, TAUCA employs a change detector, an intrinsically suitable tool for temporal analysis, which sensitively describes the changing trend in time domain. In the change detector, small shifts in probabilistic model are not ignored but accumulated until they exceed a certain range. TAUCA detects suspicious time intervals in which attacks are highly likely present. Furthermore, TAUCA reduces false alarm rate by analyzing correlation among suspicious users.
The performance of TAUCA and two other schemes are evaluated based on real user attack data collected through a cyber competition. TAUCA demonstrates significant advantages in terms of improving detection rate and reducing false alarm rate in the detection of malicious users. This work was presented at IEEE SocialCom 2010 and won the best paper award.