Research Article
Gathering Pattern Mining Method Based on Trajectory Data Stream
@INPROCEEDINGS{10.1007/978-3-030-21373-2_56, author={Ying Xia and Lian Diao and Xu Zhang and Hae-young Bae}, title={Gathering Pattern Mining Method Based on Trajectory Data Stream}, proceedings={Security and Privacy in New Computing Environments. Second EAI International Conference, SPNCE 2019, Tianjin, China, April 13--14, 2019, Proceedings}, proceedings_a={SPNCE}, year={2019}, month={6}, keywords={Gathering pattern Trajectory data streams Clustering Crowed Sliding time window}, doi={10.1007/978-3-030-21373-2_56} }- Ying Xia
Lian Diao
Xu Zhang
Hae-young Bae
Year: 2019
Gathering Pattern Mining Method Based on Trajectory Data Stream
SPNCE
Springer
DOI: 10.1007/978-3-030-21373-2_56
Abstract
Moving object gathering pattern refers to a group of incident or case that are involved large congregation of moving objects. Mining the moving object gathering pattern in massive and dynamic trajectory data streams can timely discover the anomalies in the group moving model. This paper proposes a moving object gathering pattern mining method based on trajectory data stream, which consists of two stages: clustering and crowed mining. In the clustering stage, the MR-GDBSCAN clustering algorithm is proposed. It uses the grid to index moving objects and uses the grid as a clustering object and determines the center of each cluster. In the crowed mining phase, the sliding time window is used for incremental crowed mining, and the cluster center is used to calculate the distance between different clusters, thereby improving the crowed detection efficiency. Experiments show that the proposed moving object gathering pattern mining method has good efficiency and stability.
