Abstract
The widespread use of mobile devices and social network services has made optimal location queries an important research topic. Previous studies have focused on the problem of maximum influential (Max-inf) location selection, that is, finding a location that can attract as many clients as possible. The location information of each client should be collected to process such a query. However, client location is considered sensitive information. Therefore, a privacy protection technique should be applied to Max-inf problems. Motivated by this, we propose a Max-inf problem query-processing technique with differentially private client location information. Furthermore, we present a Voronoi region-based technique to guarantee query accuracy and a Voronoi envelope-based pruning heuristic to improve query performance.
Highlights
During the past decades, a vast amount of geo-spatial data has been collected by various location-based services owing to the widespread use of mobile devices
We present a pruning technique called the Voronoi envelope filtering method (VEM), which precomputes the upper-bound of the noisy count of influece regions to reduce search space
The Voronoi region-partitioning method (VPM) and the VEM outperform the other methods in terms of accuracy and mean absolute error (MAE), but the VPM still suffers from query-processing time
Summary
A vast amount of geo-spatial data has been collected by various location-based services owing to the widespread use of mobile devices. Previous studies of geo-spatial data have focused on the maximum influential (Max-inf) location selection problem, that is, finding a location that can attract as many clients as possible. These applications generally assume that customers trust data analysts and agree to the collection of their location information without any restrictions. Customers provide information only to trusted service providers rather than data analysts To remedy this problem, we present novel Max-inf problem query-processing techniques while applying differential privacy to client location data.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have