Clustering Subtrajectories of Moving Objects Based on a Distance Metric with Multi-dimensional Weights
2014 Sixth International Symposium on Parallel Architectures, Algorithms and Programming (PAAP) (2014)
July 13, 2014 to July 15, 2014
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/PAAP.2014.59
Mining spatio-temporal data has recently gained great interest due to the integration of wireless communications and positioning technologies. Although clustering spatio-temporal data as a popular mining task has been well studied, the problem properly defining the distance between the objects to make the clustering results suit the application needs still remain largely unsolved. In this paper, for the purpose for trajectory data processing, we propose an improved trajectory segmentation algorithm and a new object distance metric that considers multiple dimensions on the characteristics of moving object's subtrajectories. Then, we use the new distance metric in a varient of the existing fuzzy clustering algorithm to improve the quality of clustering results. The experimental evaluation over real world trajectory data record with GPS demonstrates the efficiency and effectiveness of our approach.
Trajectory, Clustering algorithms, Measurement, Data mining, Vectors, Uncertainty, Global Positioning System
Y. Chen, H. Shen and H. Tian, "Clustering Subtrajectories of Moving Objects Based on a Distance Metric with Multi-dimensional Weights," 2014 Sixth International Symposium on Parallel Architectures, Algorithms and Programming (PAAP), Beijing, China, 2014, pp. 203-208.