Abstract-Driven Pattern Discovery in Databases

Abstract

In this paper, we study the problem of discovering interesting patterns in large volumes ofdata. Patterns can be expressed in user-defined terms and not only in terms of the databaseschema. The user-defined terminology is stored in a data dictionary that maps it into thelanguage of the database schema. We define a pattern as a deductive rule expressed in user-definedterms that has a degree of certainty associated with it. We present methods of discoveringinteresting patterns based on abstracts which are summaries of the data expressed in the languageof the user.