Web Services Access Prediction Used for RecommenderSystems Based on Association Rules Mining

Abstract

With the rapid expansion of web services available online, there is anin-creasing need for intelligent tools that can recommend relevant services touserswithout relying on personal data. A promising solution lies in web data min- ing,which enables the discovery of recurring and meaningful patterns withinlarge-scale datasets. This study presents a recommendation system that leverages associationrulemining on historical web service usage data. Unlike traditional approaches, theproposed system does not require user profiles; instead, it infers recommendations based solely on the implicit relationships between services that arefre-quently accessed together. Several algorithms were applied to extract these rules, and their performancewas evaluated using standard metrics such as support, confidence, andlift. Ex-perimental results across various datasets demonstrate the effectiveness of thisapproach in generating relevant service recommendations. Nonetheless, certain challenges remain-particularly in accounting for itemfrequency andserviceweighting-which will be addressed in future work.