A New Method for Web Service Recommendation Based on QoS Prediction

Abstract

As service-oriented architecture gains in popularity and grows in popularity, Web service recommendation and composition have become more important topics for research. Accurately predicting individualized QoS recommendations for recommending web services is a difficult task because of the inconsistency of the Internet and the scarcity of information regarding QoS history. Our team suggests a new framework for QoS values’ prediction and also presents two methods for clustering, User_BC and Service_BC, to support QoS prediction accuracy. Hierarchical clustering is used, based on the QoS dataset of PlanetLab1 (that) contains 200 service-user response time values, with 1,597 service values overall. In our research, we've found that our clustering-based methods beat other popular algorithms in detailed experimental comparisons and analyses.