A Bit-Encoding Approach for Frequent Sequence Mining

Sequential pattern mining is an important data mining problem with broad applications. And the classic algorithm for this problem is PrefixSpan, which is a projection-based, sequential pattern-growth approach. But it must scan projected database once in order to find a local frequent item, it may cost much time when the database is large. In this paper, we propose an encoding-based, sequential pattern-growth approach for efficient mining of sequential patterns, PSBE (PrefixSpan with Bit-Encoding), which can find local frequent item with little cost. In this approach, we encode the database with 0,1 bits, build encoded database, and use the framework of PrefixSpan to find out sequential patterns. We can improve the PrefixSpan by the property of encoding. Our experiment shows that PSBE outperforms the PrefixSpan algorithm.