Profiling Online Social Behaviors for Compromised Account Detection

Abstract

Account breakout is a serious threat to social networking users (OSN). While unsolicited spammers benefit from the trust that exists between account owners and their friends in spreading spam effectively, timely discovery of compromised accounts is a major challenge given the trust that exists between service providers, account holders and their friends. In this paper, we study the social behaviors of OSN users, that is, their use of OSN services, and their application in detecting compromised accounts. In particular, we suggest a set of social behavioral features that can effectively distinguish user social activities on the OSN. We investigate the effectiveness of these behavioral features by collecting and analyzing real user clicks to the OSN Web site. Based on our measurement study, we put a behavioral profile of the individual user by incorporating their behavioral benchmarks. The social behavior profile accurately reflects the user's OSN activity patterns. Although the original owner complies with the account's social behavior profile in a forced manner, it is difficult and costly for the fraudsters to pretend. We evaluate the ability of social behavioral profiles to distinguish different OSN users, and our experimental results show that social behavioral profiles can accurately distinguish individual OSN users and reveal hacked accounts. Indexing conditions - online social behavior, privacy, data analysis, discovery of hacked accounts.