Research On Algorithms For Mining Gathering Patterns From Trajectories Under Cloud Computing Environment

With the maturity of global positioning technology and wireless communication technology and the rapid development of location-based services(such as social media,map navigation,nearby recommendation,etc.),massive trajectories are generated to record the location sequence and activity information of mobile objects,such as spatiotemporal trajectories and semantic trajectories.These trajectories contain the movement rules and behavior patterns of moving objects,and the analysis and mining of these data can support the public event detection,public transportation optimization and other applications.This thesis studies the gathering pattern mining algorithm and implements it under cloud computing environment.The main contributions of this thesis are as follows:1.Propose algorithm CCJMGP(Cluster Containment Join based algorithm for Mining Gathering Pattering from Spatio-Temporal Trajectories)which used to mine gathering pattern by spatiotemporal trajectories.Firstly,this thesis achieve the clusters of moving objects by clustering.Then,this thesis identify the clusters with scale increasing by joining as candidate gathering patterns.Finally,this thesis verify the correctness of candidate pattern.The algorithm is oriented to spatiotemporal trajectories,focusing on the formation process of gathering groups,using the key operation-cluster containment join to identify gathering groups.In order to meet the needs of massive trajectories mining,this thesis parallelize the algorithm CCJGPM in this thesis.Parallelization algorithm divides mobile object clusters into different nodes,and implements join on multiple nodes at the same time to improve the efficiency of gathering pattern mining.Experiments with real trajectories set show that the proposed algorithm is efficient and the result is accurate.2.Propose algorithm MGPST(Mining Gathering Pattering from Semantic Trajectories)which used to mine gathering patterns by semantic trajectories.Aiming at the lack of information describing the activities of moving objects in spatiotemporal trajectories,fusing the spatiotemporal trajectories with the semantic information data to generate the semantic trajectories.The algorithm is oriented to semantic trajectories,focusing on the final state of gathering groups,mining patterns in the semantic dimension,spatial dimension and temporal dimension respectively.The results meet semantic consistency,spatial compactness and temporal persistence.In order to meet the needs of massive trajectories mining,this thesis parallelize the algorithm MGPST.Parallelization algorithm distributes moving object clusters with semantic consistency to the same node,and identifies gathering patterns at the same time by different nodes to improve the efficiency of aggregation pattern mining.Experiments with real trajectories set show that the above algorithm is more efficient than CCJGPM algorithm and the mining results have better interpretability.3.Develop prototype system of gathering pattern mining.The purpose is to display the results of gathering patterns and support the analysis and research work.The system consists of four modules:data preprocessing,pattern mining and pattern visualization.The preprocessing module is responsible for spatiotemporal trajectories preprocessing and semantic trajectories generation;the pattern mining module uses gathering pattern mining algorithm to mine gathering patterns from trajectories;and the pattern visualization module uses open source maps and chart controls to display the information of mining results.This thesis use the real trajectories set to test the system,and the results show that the system can effectively mine the gathering patterns in real world and display them in multi-dimension.