Research Article

A Shilling Group Detection Framework Based on Deep Learning Techniques

Table 1

Summary of threat models.

Threat modelISIFIT
ItemsRatingsItemsRatingsRating

GSAGenl RanNullItem is randomly chosen and rated by at most two attackersSystem meanrmax or rmin

GSAGenl AvgNullItem is randomly chosen and rated by at most two attackersItem meanrmax or rmin

GSAGenl AoPNullItem is chosen randomly from x% popular items and rated by at most two attackersItem meanrmax or rmin

GSAGenl PIAItems are randomly chosen from power items and rated by less than three attackersItem meanNullrmax or rmin

GSAGenl GOATItems are randomly chosen from popular items rated by active users and rated by less than three attackersLearned by GANItems are randomly chosen from popular items rated by active users and rated by less than three attackersLearned by GANrmaxāˆ’1