Reputation-based Service Composition

Cloud computing is the latest evolution in the field of distributed computing. This advocates the composition of basic distributed web services to form a more complex Second Generation Service encapsulating its complete business workflow. Trust and reputation management have emerged in industry and academia for selecting the best web services available and thus forming the best-in-class compositions. This paper focuses on the state-of-art for managing reputation and how factors involving reputation can be used to construct well-defined compositions. I surveyed some techniques, presented in literature, involving different implementations of reputation managers and their integration with the web service architectures. Every technique is analysed with their advantages and disadvantages and a comparison with other methods, is stated. Finally, I also present On-the-Fly Computing, a research project at the University of Paderborn and the reputation matching approach used there to automatically compose services published in a dynamic market. I also perform a comparison of On-the-Fly Computing reputation technique with other techniques discussed. It can be concluded that the reputation approaches have a major dependency on the domain and technologies used. Hence, there is no single technique which can be termed as the best-in-class.