A mixed opinion-based reputation prediction approach for reliable web service discovery

Abstract

Prediction of service reputation is an important research issue to include reliable services in service discovery and composition. A user of the service plays an important role in the prediction of service reputation. Therefore, feedback ratings provided by service users cannot be completely avoided. Most of the existing approaches fully trust on service users' feedback rating for reputation prediction, which often leads to bias towards positive or negative ratings. In this paper, a mixed opinion-based reputation prediction approach is proposed, which uses theoretical t-probability distribution and propositional logic to assess and predict the service reputation. The proposed approach combines users' feedback ratings and run time access rate to minimise the probable bias towards feedback ratings provided by users. Experimental result shows that proposed approach provides effective solution for prediction of service reputation, which can be helpful in performing reliable service discovery and composition.