Discovery of Unexpected Fuzzy Recurrence Behaviors in Sequence Databases

Abstract

The discovery of unexpected behaviors in databases is an interesting problem for many real-world applications. In previous studies, unexpected behaviors are primarily addressed within the context of patterns, association rules, or sequences. In this paper, we study the unexpectedness with respect to the fuzzy recurrence behaviors contained in sequence databases. We first propose the notion of fuzzy recurrence rule, and then present the problem of mining unexpected sequences that contradict prior fuzzy recurrence rules. We also develop, UFR, an algorithm for discovering the sequences containing unexpected recurrence behaviors. The proposed approach is evaluated with Web access log data.