Mining weighted sequential patterns in a sequence database with itemset-interval measurement

Abstract

Weighted sequential pattern mining algorithms discover weighted sequences with considering the different significance of each item in a sequence database. But current algorithms have not considered the importance of the itemset-intervals information between the two items in a same itemset. Hence, although a large number of sequences had been discovered, most of them are not useful for analysis. In this study, we propose a new algorithm, called ItemSet-interval Weighted Sequences (ISiWS), to solve the problem about efficient discovering useful sequences. In ISiWS, a matrix structure, called Transaction Bit Matrix (TBM), represents a sequence. ISiWS first uses TBMs to represent the sequences in a sequence database. Then, it utilizes projected technology to discover weighted sequences, and an approximate sequence match algorithm is applied to calculate support of sequences based on their itemset-intervals. Experiments show that ISiWS produces a significantly less number of weighted sequences than those of WSpan. © 2014 Springer International Publishing.