Towards privacy-preserving category-aware POI recommendation over encrypted LBSN data

With the popularity of location-based social networks (LBSNs), locations and social relationships, point-of-interests (POIs) categories have been considered to be important factors in POI recommendation services. The boom of cloud computing has driven data owners to outsource their LBSN data and services to the cloud. However, as the data usually contains sensitive information, it should be encrypted before being outsourced. Consequently, the POI recommendations have to be processed over encrypted data. Although several secure LBSNs-based POI recommendation schemes have been proposed, they either do not apply to the outsourcing scenario or have issues with recommendation accuracy. To overcome these limitations, we propose an efficient and privacy-preserving category-aware POI recommendation scheme (PCAPR) over encrypted LBSN data. Specifically, we first index social relationships with Bloom filters and organize them into a vantage point tree. Then, we design a Merge-tree to index the user-POI dataset. Based on these two trees, we design an efficient LBSNs-based and category-aware POI recommendation algorithm to support a threshold POI recommendation. After that, we design two secure protocols to protect the privacy in the designed algorithm and present a PCAPR scheme. Our analysis confirms the PCAPR scheme ensures privacy while our performance evaluation underscores its efficiency. •A novel category-aware POI recommendation scheme, called PCAPR, is proposed.•PCAPR comprehensively considers locations, social relationships, and POI categories.•PCAPR is secure for recommendation request and location-based social network data.•PCAPR can recommend POIs that are close, visited by neighbors, and have specified categories.•PCAPR is efficient in terms of computational cost and communication overhead.