A Study on Post mining of Association Rules Targeting User Interest

Association Rule Mining means discovering interesting patterns with in large databases. Association rules are used in many application areas such as market base analysis, web log analysis, protein substructures. Several post processing methods were developed to reduce the number of rules using nonredundant rules or pruning techniques such as pruning, summarizing, grouping or visualization based on statistical information in the database. As such, problem of identifying interest rules remind the same. Methods such as Rule deductive method, Stream Mill Miner (SMM), a DSMS (Data Stream Management Systems), Medoid clustering technique (PAM: Partitioning around medoids), Constraint-based Multi-level Association Rules with an ontology support were developed but are not effective. The number of rules generated by Apriori, FPgrowth depends on statistical measures such as support, confidence and may not suit the requirements of user. Methods that use ranking algorithm and IRF (Item Relatedness Filter) have the drawbacks of using filters during pruning stage. The paper studies methods that were proposed for post processing of association rules and proposes a new method for extracting association rules based on user interest using MIRO (Mining Interest Rules Using Ontologies) framework that uses correlation measures combined with domain ontology, succint constraints. KeywordsAssociation Rules, Association Rule Mining, Ontology, correlation measures, user constraints.