Ready for emerging threats to recommender systems? A graph convolution-based generative shilling attack

To explore the robustness of recommender systems , researchers have proposed various shilling attack models and analyzed their adverse effects. Primitive attacks are highly feasible but less effective due to simplistic handcrafted rules, while upgraded more powerful costly difficult deploy because they require knowledge from recommendations. In this paper, we a novel called Graph cOnvolution-based generative ATtack (GOAT) balance attacks’ feasibility effectiveness. GOAT adopts primitive paradigm that assigns items for fake users by sampling generates ratings deep learning-based model. It deploys adversarial network (GAN) learns real rating distribution generate ratings. Additionally, generator combines tailored graph convolution structure leverages correlations between co-rated smoothen enhance authenticity. The extensive experiments on two public datasets evaluate GOAT’s performance multiple perspectives. Our study demonstrates technical building intelligent model with much-reduced cost, enables analysis threat such an guides investigating necessary prevention measures.