Multi-relational Sequence Pattern Mining Method Based on Improved Prefix Tree in the Star Model

With the development of information technology and the increasing amount of data, the way of storing data in single table can not meet the actual needs, it will highlight the importance of the research on multi-relational sequence mining. This paper presents a multi-relational sequence pattern mining algorithm using the variant prefix tree, and the frequent sequence pattern is obtained by connecting all the tables in the improved star model. Using discretization method, combined with users´ specified information, as well as the improved structure and the chi-square test of the prefix tree pruning strategy, the sequence patterns can reflect different relationships between entities, providing the effective solution to cross-links between tables in mining issues that the single-table mining failed. The experiments show that the proposed algorithm can efficiently mining the multi-relational sequence patterns with a good performance.