Research On Mining Spatiotemporal Co-occurrence Paterns From Large Data Sets

As the volume of spatiotemporal data continues to increase significantly due to both the growth of database archives and the increasing number of spatiotemporal collection equipment, automatic and semi-automatic pattern analysis becomes more essential. It is meaningful and challenging for us to extract interesting patterns from these large spatiotemporal data sets. Co-occurrence Pattern represents subsets of different object-types whose instances are co-located together for a signification fraction of time.In this paper, our aim is to mine co-occurrence patterns from large spatiotemporal data set. According to the existing co-occurrence pattern mining algorithms, we make research and exploration in order to improve the calculation efficiency of mining co-occurrence patterns and also solve the problem of analysis large spatiotemporal data sets.The main research work are as follows:(1) Improve the calculation efficiency of mining co-occurrence patterns. In order to improve the calculation efficiency of mining co-occurrence patterns, this paper proposes a novel co-occurrence pattern mining algorithm MDCOP Graph Miner by using Time Aggregated Graph (TAG). MDCOP Graph Miner algorithm reduces the computation time and improves the efficiency of mining co-occurrence patterns.(2) Analysis large spatiotemporal data sets. In order to solve the problem of mining co-occurrence patterns from large spatiotemporal data sets, this paper presents LDMDCOP Graph Miner algorithm.By using file to store MDCOP Graph, we can solve data storage problem;Meanwhile we create index of the MDCOP Graph file for the purpose of improve the efficiency of querying.Through the methords above,we solved the problem of mining co-occurrence patterns from large spatiotemporal data sets.