Fast and Efficient Technique for Mining Frequent Closed Sequences

Frequent patter mining has been around ever since the data mining domain came into popularity. However, there was an argument in this domain that one should min only closed frequent patterns instead of mining all frequent patterns. The reason being that the former gives rise to very compact and efficient result set. The drawback of most of the existing closed sequence mining algorithms is that they rely on an approach known as candidate maintenance-and-test. This approach is costly in terms of memory usage and the time taken. To overcome these drawbacks, this paper presents a new algorithm that can effectively mine closed frequent sequence patterns without the need for maintaining candidates. It also prunes search space for reducing cost and time of the operations. Extensive experiments are made with many real world datasets. The experimental results reveal that the proposed algorithm and pruning techniques are efficient