IPACT: Improved Web Page Recommendation System Using Profile Aggregation Based On Clustering of Transactions

Problem statement: Recently, Web usage mining techniques have been widely used to build recommendation systems especially for anonymous users. Approach: Assigning the current user to the best web navigation profile with similar navigation activities will improve the ability of the prediction engine to produce a recommendation list then introduce it to the user. This study presents iPACT an improved recommendation system using Profile Aggregation based on Clustering of Transactions (PACT). Results: iPACT shows better prediction accuracy than the previous methods PACT and Hypergraph. Conclusion: The usersʹ interests change over time; hence an incremental and adaptive web navigation profiling is a key feature for the future works.