2015/06/01 by Muhammad Ikram, Ikram, Muhammad, Lucky Onwuzurike +13
Computer Science · Social Sciences · #Cryptography and Security (cs.CR) #Cybercrime and Law Enforcement Studies #FOS: Computer and information sciences #Misinformation and Its Impacts #Social and Information Networks (cs.SI) #Spam and Phishing Detection
paper · pdf · doi:10.48550/arxiv.1506.00506
openalex publication_date 2015/06/01 · openalex created_date 2022/09/29 · openalex updated_date 2026/07/28
As businesses increasingly rely on social networking sites to engage with\ntheir customers, it is crucial to understand and counter reputation\nmanipulation activities, including fraudulently boosting the number of Facebook\npage likes using like farms. To this end, several fraud detection algorithms\nhave been proposed and some deployed by Facebook that use graph co-clustering\nto distinguish between genuine likes and those generated by farm-controlled\nprofiles. However, as we show in this paper, these tools do not work well with\nstealthy farms whose users spread likes over longer timespans and like popular\npages, aiming to mimic regular users. We present an empirical analysis of the\ngraph-based detection tools used by Facebook and highlight their shortcomings\nagainst more sophisticated farms. Next, we focus on characterizing content\ngenerated by social networks accounts on their timelines, as an indicator of\ngenuine versus fake social activity. We analyze a wide range of features\nextracted from timeline posts, which we group into two main classes: lexical\nand non-lexical. We postulate and verify that like farm accounts tend to often\nre-share content, use fewer words and poorer vocabulary, and more often\ngenerate duplicate comments and likes compared to normal users. We extract\nrelevant lexical and non-lexical features and and use them to build a\nclassifier to detect like farms accounts, achieving significantly higher\naccuracy, namely, at least 99% precision and 93% recall.\n