A Generic Data Synthesis Framework for Privacy-Preserving Point-of-Interest Recommender Systems
Longyin Cui, Xiwei Wang, Ting Gu · 2023
Personalization services offered by Point-of-Interest (POI) recommender systems are becoming increasingly popular, especially in the context of mobile devices. However, data privacy regulations and user concerns regarding privacy often prevent the transfer and storage of user data, which poses a challenge for these systems. To address this issue, privacy-preserving recommender systems have gained importance. This paper proposes a generic framework for generating synthetic user data for POI recommendations based on differential privacy, random response, and user grouping. The proposed framework can accommodate various data feedback without compromising privacy and is compatible with non-private recommender systems, allowing for future improvements and flexibility. Our experiments on real-world datasets demonstrate that the framework strikes a balance between privacy protection and accurate recommendations.