TP-mine:基于分割簇的RFID轨迹数据增量聚类算法

TP-mine:A partition cluster based incremental clustering algorithm for RFID trajectory data mining

针对具有增量特性的射频识别(RFID)轨迹数据的挖掘进行了研究,提出了轨迹聚类算法TP-mine。该算法将每个新的轨迹简化成一个有向线性片段以便于找到轨迹子部分的聚类,使用分割簇来存储紧密的相似轨迹线性片段,比原始的轨迹占的空间要小。TP-mine算法在整体聚类时对分割聚类所生成的分割簇进行操作,而不是对所有时间段的全部轨迹进行操作,可以高效地得出轨迹的聚类结果,从而实现对RFID技术所产生的轨迹数据进行有效挖掘,来发现移动对象潜在的移动趋势。

Aiming at the mining of the RFID （ radio frequency identification） trajectory data with the incremental properties, the study put forward a trajectory clustering algorithm, TP-mine. The TP-time simplifies each new trajectory into a directed linear segment in order to find the clusters of the trajectory＇ s subparts, to store the similar trajectory line segments in partition-clusters which take much smaller space than raw trajectories. The integrity-clustering is performed on the set of partition-clusters rather than on all trajectories over the whole time span. Thus the integrity- clusters are generated efficiently to show the clustering results of trajectories. So the RFID trajectory data can be mined and the moving objects＇ potentially moving trends can he fnund