A Hybrid Collaborative Filtering Technique for Web Service Recommendation using Contextual Attributes of Web Services

Abstract

Quality of Service (QoS) Aware recommender system considers the quality of service to recommend personalized web services to the user. Quality of Service parameters also includes response time and throughput that a user receives when invoking a web service. There exist numerous collaborative filtering techniques that tend to predict Quality of Service value; however, existing techniques only use the client-side information of QoS and neglect the service's contextual attributes. This paper proposes a new Web Service Recommendation System that will consider the contextual attributes of Web services. The proposed method collects the contextual properties from WSDL files to cluster Web services based on their attribute similarities. Thus, more accurate neighbour selection takes place and prediction value is determined using QoS record; in addition to this, a user-influenced prediction value is also determined. To map both, service, and user influence on QoS prediction, a hybrid memory-based CF model is developed. The effectiveness and reliability of the proposed system is verified by the results of experiments.