Efficient Parallel Processing of Semantic Trajectory Similarity Joins

Matching similar pairs of trajectories, called tra-jectory similarity join, is a fundamental functionality forthe Internet of Everything (IoE). We obverse that keyword-augmented trajectories are becoming increasingly popular. Inthis light, we investigate semantic trajectory similarity (STS)join that consists of two subproblems, threshold-based STSJoin and top-kSTS(k-STS) Join. Each semantic trajectory isa sequence of geo-textual objects with both location and textinformation. Specifically, given two sets of semantic trajectoriesand a threshold theta or result numberk, the STS Join returnsall pairs of semantic trajectories from the two sets with spatio-textual similarity no less than theta,andthek-STS Join returnskmost similar pairs of semantic trajectories from the two sets.To enable efficient STS andk-STS Joins processing on largesets of semantic trajectories, we present a two-phase parallelsearch algorithm. We first group semantic trajectories basedon their text information. The algorithm's per-group searchesare independent of each other and thus can be performedin parallel. We generate spatial and textual summaries foreach trajectory batch and develop batch filtering techniques toprune unqualified trajectory pairs in a batch mode. Next, wepropose a divide-and-conquer algorithm to derive bounds ofspatial similarity and textual similarity between two semantictrajectories, which enable us filter out dissimilar trajectory pairsefficiently. Further, hierarchical batch filtering join algorithmis developed to processk-STS Join. Experimental study withlarge semantic trajectory data confirms that our algorithm ofprocessing semantic trajectory join is capable of substantiallyoutperforming well-designed baselines.