SemTraClus: an algorithm for clustering and prioritizing semantic regions of spatio-temporal trajectories

摘要

The widespread acceptance of context-sensing applications is generating voluminous movement data on high speed, which has fueled research studies in mining of spatio-temporal trajectories. The analysis of space-time points in trajectory gives insightful knowledge on the pattern of the mobility of the object and on the interest evinced by visitors in a geographic location. Significant locations of a geographical area, called Points of Interests, are extracted by means of spatial and temporal features of the moving object and enriching them with semantic information is a new trend in spatio-temporal data mining. In this paper, an algorithm called SemTraClus is proposed for identifying and clustering the semantic subtrajectories of moving traces of multiple objects. The semantic regions are clustered using the DBSCAN method. Finally, it generates a Weightage Participation value which provides priorities of user interest in different semantic cluster regions. It also identifies the most representative user trajectory that has traveled through relevant locations. To the best of our knowledge, this is the first work that clusters multiple trajectories for the identification of semantic points, considering spatial and temporal features simultaneously and providing prioritized location list. Experiments show that the proposed algorithm achieved good results in identifying significant locations and prioritizing it.