Simulating real profiles for shilling attacks: A generative approach

作者：

Highlights：

• We propose Variational Autoencoder to replicate real profiles distribution.

• Generated profiles are converted to malicious by adding an intent.

• Our model outperforms Shilling Attack models on model-based CF techniques.

摘要

•We propose Variational Autoencoder to replicate real profiles distribution.•Generated profiles are converted to malicious by adding an intent.•Our model outperforms Shilling Attack models on model-based CF techniques.

论文关键词：Recommender systems,Collaborative Filtering,Shilling Attack,Generative Model,Variational Autoencoder

论文评审过程：Received 18 September 2020, Revised 1 June 2021, Accepted 9 August 2021, Available online 16 August 2021, Version of Record 23 August 2021.

论文官网地址：https://doi.org/10.1016/j.knosys.2021.107390