Mining abnormal path from RFID path data sets
Dong Guobin, Anrong Xue, Zhao Baotong · Jisuanji yingyong yanjiu · 2013
Radio frequency identification(RFID) technology is becoming a prevalent tool in tracking commodities in supply chain management application.The movement of commodities through the supply chain forms a gigantic path data.Each node in the path data contains location information and time information,so that the path data is more complex than the general sequence of data.Existing sequence data outlier detection algorithms are not suitable for processing path data.According to the characteristics of the path data,this paper proposed extended probabilistic suffix tree(EPST) model to detect abnormal path.The similarity of a path and path data sets could be efficiently calculated with the EPST.The path data had a property of short memory and this could be used to simplify the calculation of similarity.The advantage of EPST was that when calculating the similarity it took into account the impact of the time and location information.The experiments show that the proposed algorithm can accurately detect the abnormal path and has a lower space complexity.