Efficient Mining of Regional Semantic Patterns from Semantic Trajectories on Cloud Computing

Semantic trajectory pattern mining is becoming more and more important with the efficient semantic enrichment process methods, which is the process of semantic annotation for spatiotemporal trajectories. Existing works on semantic trajectory pattern mining focus on sequential patterns that occur frequently globally in the semantic trajectory data, ignoring the ones that are locally significant within a region, which is called regional semantic pattern. As a consequence, the existing regional semantic pattern mining algorithm RegMiner is time consuming. Based on this motivation, we propose an improved efficient regional semantic pattern mining algorithm called MRSP, which consists of three parts: candidate semantic category sequence discovery, regional semantic pattern recognition, and parallelization. The innovations of MRSP lie in that we use an improved neighborhood pruning algorithm (NPSP) and an improved neighborhood query algorithm (NSSI) in candidate semantic category sequence discovery mentioned above, and adopt a parallelization algorithm to do calculation based on Hadoop platform. Finally, we evaluate our approach on real dataset in Shanghai. Compared with RegMiner, MRSP has the same effect in mining regional semantic patterns, and the efficiency is greatly improved.