Privacy-preserving multi-criteria collaborative filtering

作者：

Highlights：

• Randomized perturbation-based data disguising procedure is applied to multi-criteria user preference data.

• Conventional perturbation-based random filling protocols were integrated into the multi-criteria preference domain.

• A novel entropy-based privacy-preserving MCCF framework is proposed which allows controlling of privacy-specific parameters dynamically for each user and sub-criteria individually.

• Proposed novel scheme significantly improves prediction accuracy while maintaining an identical level of privacy with conventional privacy-preserving scenario.

摘要

•Randomized perturbation-based data disguising procedure is applied to multi-criteria user preference data.•Conventional perturbation-based random filling protocols were integrated into the multi-criteria preference domain.•A novel entropy-based privacy-preserving MCCF framework is proposed which allows controlling of privacy-specific parameters dynamically for each user and sub-criteria individually.•Proposed novel scheme significantly improves prediction accuracy while maintaining an identical level of privacy with conventional privacy-preserving scenario.

论文关键词：Collaborative filtering,Entropy,Multi-criteria,Recommender systems,Privacy

论文评审过程：Received 13 September 2018, Revised 9 February 2019, Accepted 10 February 2019, Available online 20 February 2019, Version of Record 20 February 2019.

论文官网地址：https://doi.org/10.1016/j.ipm.2019.02.009