A Novel Approach for Efficient Computation of Community Aware Ridesharing Groups

The evolution of ridesharing services has reduced the road traffic congestions in recent years. However, a major concern for ridesharing services is sharing rides with strangers. To address this issue, a few ridesharing approaches have considered social closeness of group members for identifying a ridesharing group. Again, users do not feel comfortable to disclose such personal data (e.g, friendship information) with an untrusted service provider for privacy reasons. We propose a novel way to form ridesharing groups that reveals user social data in community levels, and ensures that a group member shares at least k common communities with at least other m members in the ridesharing group, where k and m are personalized parameters of every group member. We formulate a Community aware Ridesharing Group (CaRG) query that satisfies the constraints of m and k, and returns a ridesharing group with the minimum cost in terms of the spatial proximity of riders from the driver. We show in experiments that our approach to process CaRG queries outperforms a baseline approach with a large margin.