Trajectory Query Based on Trajectory Segments with Activities

Abstract

Searching trajectories with activities has attracted much attention in the last decade. Existing studies tend to find trajectories with activities matched to the required keywords. However, returned trajectories may have a satisfying textual matching but are spatially far from query locations. In this paper, differing with traditional work which return entire trajectories without combination, we focus on the intersecting trajectory segments and combine them into a new trajectory. A challenge of this problem is how to find qualified trajectory segments from the large search space and combine them into required trajectories. To this end, we organize trajectories into a hybrid index which enables us to utilize spatial information to prune search space efficiently. In addition, we propose a algorithm to search intersecting trajectory segments and combine them into qualified trajectories according to requirements. The effectiveness of our method is verified by empirical studies based on a real trajectory data set and a synthetic data set.