Scalable Trajectory Similarity Search Based on Locations in Spatial Networks

In this paper, we propose an efficient query processing algorithm that returns the trajectory results in a progressive manner. We limit the calculation of pairwise shortest path distances between the set of query locations and the spatial nodes, by highly reducing the preprocessing requirements. Also, we introduce a spatiotemporal similarity measure, based on which the temporal-to-spatial significance of the trajectory results can be easily modified and the query locations can be spatially prioritized according to users' preferences. In our experiments with a real-world road network, we show that the proposed method has approximately ten times less preprocessing requirements than the competitive methods and reduces the search time by two orders of magnitude at least.