Efficient Processing of Probabilistic Group Nearest Neighbor Query on Uncertain Data

Uncertain data are inherent in various applications, and group nearest neighbor (GNN) query is widely used in many fields. Existing work for answering probabilistic GNN (PGNN) query on uncertain data are inefficient for the irregular shapes of uncertain regions. In this paper, we propose two pruning algorithms for efficiently processing PGNN query which are not sensitive to the shapes of uncertain regions. The spatial pruning algorithm utilizes the centroid point to efficiently filter out objects in consideration of their spatial locations; the probabilistic pruning algorithm derives more tighter bounds by partitioning uncertain objects. Furthermore, we propose a space partitioning structure in order to facilitate the partitioning process. Extensive experiments using both real and synthetic data show that our algorithms are not sensitive to the shapes of uncertain regions, and outperform the existing work by about 2-3 times under various settings.