Combining clustering with moving sequential pattern mining: A novel and efficient technique

Abstract

Sequential pattern mining is a well-studied problem. In the context of mobile computing, moving sequential patterns that reflects the moving behavior of mobile users attracted researchers’ interests recently. In this paper a novel and efficient technique is proposed to mine moving sequential patterns. Firstly the idea of clustering is introduced to process the original moving histories into moving sequences as a preprocessing step. Then an efficient algorithm called PrefixTree is presented to mine the moving sequences. Performance study shows that PrefixTree outperforms LM algorithm, which is revised to mine moving sequences, in mining large moving sequence databases.