GSLDA: LDA-based group spamming detection in product reviews

摘要: Online product reviews are becoming increasingly important due to their guidance function in people’s purchase decisions. As being highly subjective, online subject opinion spamming, i.e., fraudsters write fake or give unfair ratings promote demote target products. Although there have been much efforts this field, the problem is still left open difficulties gathering ground-truth data. more and people using Internet everyday life, group review which involves a of writing hype-reviews (promote) defaming-reviews (demote) for one products, becomes main form spamming. In paper, we propose LDA-based computing framework, namely GSLDA, spamming detection completely unsupervised approach, GSLDA works two phases. It first adapts LDA (Latent Dirichlet Allocation) context order bound closely related spammers into small-sized reviewer cluster, then it extracts high suspicious groups from each LDA-clusters. Experiments on three real-world datasets show that can detect quality spammer groups, outperforming many state-of-the-art baselines terms accuracy.