Recognizing unexpected recurrence behaviors with fuzzy measures in sequence databases
Dong Li, Anne Laurent, Pascal Poncelet · 2008
The recognition of unexpected behaviors in databases is an important problem in many real-world applications. In the previous studies, the unexpectedness is mainly stated within the context of the most-studied patterns, association rules, or sequential patterns. In this paper, we first propose the notion of fuzzy recurrence rule, a new kind of rule-based behavior in sequence databases, and then we introduce the problem of recognizing unexpected sequences contradicting the beliefs on fuzzy recurrence rules, with fuzzy measures. We also develop, UFR, an algorithm for discovering unexpected recurrence behaviors in a sequence database. Our approach is evaluated with Web access log data.