No abstract

No abstract

Abstract. Functional, nonfunctional, just-in-time approaches to composing web services span the sub-disciplines of software engineering, data management, and artificial intelligence. Our research addresses the process that must occur once the composition has completed and stakeholders must investigate historical and online operations/data flow to reengineer the process either offline or in real-time. This research introduces an effective reference model to assess the message flow of long-running service workflows. We examine Dynamic Bayesian Networks (dDBNs), a data-driven modeling technique employed in machine learning, to create service workflow reference models. Unlike other reference models, this method is not limited by static assumptions. We achieve this by including the trend and time varying variables in the model. We demonstrate this method using a flight dataset collected from various airlines.