Effective decision support for semantic web service selection
The objective of this dissertation is to demonstrate the feasibility of the vision of the Internet of Services based on Semantic Web Services by suggesting an approach to end-user mediated Semantic Web Service selection. Our main contribution is an incremental and interactive approach to requirements elicitation and service selection that is inspired by example critiquing recommender systems. It alternates phases of intermediate service recommendation and phases of informal requirements specification. During that process, the user incrementally develops his service requirements and preferences and finally makes a selection decision. We demonstrate how the requirements elicitation and service selection process can be directed and focused to effectively reduce the system's uncertainty about the user's service requirements and thus to contribute to the efficiency of the service selection process. To acquire information about the actual performance of available services and thus about the risk that is associated with their execution, we propose a flexible feedback system, that leverages reported consumer experiences made in past service interactions. In particular, we provide means that allow to detailedly describe a service's performance with respect to its multiple facets. This is supplemented by a user-adaptive method that effectively assists service consumers in providing such feedback as well as a privacy-preserving technique for feedback propagation. We also demonstrate that available consumer feedback can be effectively exploited to assess the degree and kind of risk that is associated with the execution of an offered service and show how the user can be effectively made aware of this risk. In contrast to many other approaches related to Semantic Web Service technology, we performed an extensive and thorough evaluation of our contribution and documented its results. These show the effectiveness and efficiency of our approach.