Spotting Collective Behaviour of Online Frauds in Customer Reviews.

摘要: Online reviews play a crucial role in deciding the quality before purchasing any product. Unfortunately, spammers often take advantage of online review forums by writing fraud to promote/demote certain products. It may turn out be more detrimental when such collude and collectively inject spam as they can complete control users' sentiment due volume inject. Group detection is thus challenging than individual-level unclear definition group, variation inter-group dynamics, scarcity labeled group-level data, etc. Here, we propose DeFrauder, an unsupervised method detect reviewer groups. first detects candidate groups leveraging underlying product graph incorporating several behavioral signals which model multi-faceted collaboration among reviewers. then maps reviewers into embedding space assigns score each group that comprising with highly similar traits achieve high score. While comparing five baselines on four real-world datasets (two them were curated us), DeFrauder shows superior performance outperforming best baseline 17.11% higher NDCG@50 (on average) across datasets.