ARIPSO: Association rule interactive postmining Using Schemas And Ontologies

Abstract

One of the central problem in the field of knowledge discovery is the interestingness problem and the huge number of association rules being mined. As rule interestingness depends on user knowledge and goals, past research and applications have found that it's too easy to discover huge number of rules and patterns. The existing methods in literature like postprocessing and algorithms to reduce itemsets and nonredundant rules do not guarantee mining of interesting rules for the user. As, these methods depend on statistical information rule discovery becomes a tedious process. In Data Mining, the usefulness of association rules is strongly limited by the huge amount of delivered rules. This paper proposes a new interactive approach ARIPSO to prune and filter the discovered rules. We propose to integrate user knowledge in association rule mining using three types of formalism: Ontologies, Rule Schemas and interactive framework. First, we use Domain and Background Ontology with the concept of GAR (Generalized Association Rules), Rule schema is based on specification language as represented by user knowledge and the interactive framework of ARIPSO assist the user throughout the analyzing task and permits him for easy selection of rules and also to revise the information that is proposed. Moreover page ranking algorithm is used for retrieval of frequently accessed rules and the concept of privacy is enforced while mining. Thus on applying our new approach the number of rules is reduced to several dozens or less.