Geometrically-enhanced group-based mobile queries with uncertainty
An important class of community-based activities of mobile users is searching and querying for common locations to visit or come together for a specific task. For this purpose, Group Nearest-Neighbor (GNN) queries are used a generalization of nearest-neighbor queries where the goal is to find one or more points from a set of destination points that have the smallest total distance from all query points. Since users are situated at the query points as members of a group, and the perception of people about distance can be different, the classic GNN models cannot be used such differences. On the other hand, more rich and multi-faceted distance models based on type-2 fuzzy logic exist but they require heavy computations which makes them difficult to use in real-world applications. In this paper, we propose a method based on spatial tessellation and fuzzy clustering of destination points that helps to compute the approximate response to GNN query in efficient time. For this purpose, Voronoi diagrams and two fuzzy clustering methods are compared using several evaluation criteria. The results show that the proposed method provides higher performance while keeping a good quality of approximation in terms of similarity between ideal and approximated response sets. The proposed uncertainty management method also improves the robustness of system to small movements of mobile group members.