An improved prefixspan algorithm research for sequential pattern mining

Abstract

PrefixSpan, the classic sequential patterns mining algorithm, has the problem of large expenses in constructing projected databases. A Sequential Patterns Mining based on Improved PrefixSpan (SPMIP) algorithm is proposed on the basic of the defects above. This algorithm can reduce the scale of projected databases and the time of scanning projected databases through adding the pruning step and reducing the scanning of certain specific sequential patterns production. In this way, algorithm efficiency can be raised, and the sequential patterns needed are obtained. The experiment results show that the SPMIP algorithm is more efficient than the PrefixSpan algorithm while the sequential patterns obtained are not affected.