Autonomic Trust Extraction for Trustworthy Service Discovery in Urban Computing

Trustworthy service discovery against malicious services is an issue in pervasive computing where interactions between user and service occur spontaneously. For trustworthy service selection, reputation based trust model which uses user evaluation to derive the trustworthiness of service behavior is a well known approach. However, it requires voluntary user participation and often has a problem with the lack of gathered data. In general, task based computing provides a task to user which is a composition of services rather than unit services.Therefore, user can´t give direct evaluation on unit service. In this paper, we present an autonomic trust extraction method that trust value of service behavior can be extracted without any user intervention. In our method, the interaction between user and service is monitored, and trust value for a service is automatically determined after each interaction according to the number of recurrent interaction with a service and the fact whether the interacted service is bookmarked or not. We also present simulation results that evaluate the proposed model.The results show that our model captures the degree of trustworthy in service behavior r with about 70% ~ 80% of correlation between the actual trust and the estimated trust when user bookmarking ratio is about over 30%.