| Service-oriented computing provides organizations with methodologies and technologies that enable them to link multiple software systems using platform independent interfaces and contracts thereby creating a so-called Virtual Organization (VO). Unfortunately, as of today, there are challenges that can limit the implementation of a true service-oriented environment. One of the most noticeable challenges is the issue of "trust". During service-oriented process integration, participants may face uncertainties due to the nature of Web Service technologies. These uncertainties arise from the limitations of current Web Service support that is required for collaboration between participants in the VO. As a result, resolution of such limitations of Web Services is both very necessary and highly desirable.; In this thesis, we propose a new framework that we call the Web Service Collaborative Process Coordinator (WSCPC), which provides support for a reliable service-oriented collaborative environment. Using our proposed approach, partners can share their workspaces through Web Service semantics in heterogeneous processes. Behaviors in the VO can be predicted, monitored, and analyzed so as to determine if a goal can be successfully achieved. Such behaviors can also be enacted on-the-fly by partners based on the needs.; To this end, we first propose a trustworthiness model that can be used to predict performance and reliability of service behavior. The trustworthiness model employs a probabilistic method called probabilistic Latent Sematic Analysis (pLSA) that eliminates noise in the pool of service ratings. Using this methodology, service consumers can expect a particular degree of satisfaction and estimate percentage measure that predicts to what extent which they will be disappointed. Our trustworthiness model can also be integrated into system analysis methodologies such as stochastic Petri nets, and hence support run-time quantitative predictions.; Moreover, we propose several service models including a service definition model, a service registry model, and a service interaction model. These service models are described using the Ontology Web Language (OWL), an XML knowledge representation based on Description Logic (DL). As a result, any peer's capability and behavior can be logically inferred and enacted in a heterogeneous, collaborative software environment. Under these service models, collaborators in a VO can share communicational behaviors and enact partners' behavior in a flexible and platform neutral way. |
