A QoS and QoE based integrated model for bidirectional web service recommendation

Abstract

For a given requirement, identifying relevant Web services and recommending the best ones is an important task in Service-oriented application development. In this paper, a composite model that leverages Quality of Service (QoS) and Quality of Experience (QoE) for bidirectional Web service recommendation (bi-WSR) is proposed. The QoS based recommendation model is built on degree of user satisfaction, calculated using a special normalization technique and user satisfaction functions like response time and throughput. The QoE model is trained on a dataset containing positive, negative and neutral textual reviews of web services for sentiment analysis and mapped to each service's QoS values using a clustering method. This further optimizes the recommendation of web services to consumers, as the sentiment score of reviews is integrated with the user satisfaction using weighted average scoring. To describe the relationship between both web services consumers and providers consumers, a cube model is built. For recommending services to consumers and recommending potential consumers to service providers, hybrid collaborative filtering based techniques were used. The results obtained when only QoS is used, and when QoS and sentiment analysis scores are integrated to form QoE showed significant improvement in the quality of recommendation. � 2018 Pacific Neighborhood Consortium (PNC).