Web Service Recommendations Based on Time-Aware Bayesian Networks

With the increasing number of services published on the Web, it is useful to derive desirable service execution plans by identifying relevant and reliable services via an intelligent and automated recommendation process. Different from other proposals, this paper proposes to exploit time-dependant relationships among web services and their quality of service to make better recommendations. We employ time-aware Bayesian networks to reveal time dependency relationships for quality of service from service logs. Service selection is then guided using the latest time-step quality of service of related services. The short listed services can be further evaluated by business social trust paths to identify their trustworthiness. Our experiments demonstrate the effectiveness of our proposed service ranking method.