Discovery of Sequential Patterns Based On Constraint Patterns

This paper proposes a method that discovers sequential patterns corresponding to user’s interests from sequential data. This method expresses the interests as constraint patterns. The constraint patterns can define relationships among attributes of the items composing the data. The method recursively decomposes the constraint patterns into constraint subpatterns. The method evaluates the constraint subpatterns in order to efficiently discover sequential patterns satisfying the constraint patterns. Also, this paper applies the method to the sequential data composed of stock price indexes and verifies its effectiveness through comparing it with a method without using the constraint patterns. Keywords—Sequential pattern mining, Constraint pattern, Attribute constraint, Stock price indexes