Automatedweb service composition using semantic descriptions

Web services are self-describing, platform independent applications that can be accessed over the Internet. Web services enable us to dynamically find, bind and consume the needed functionality. Semantic Web services are being developed with the vision of automating this process and to facilitate interoperable machine to machine interaction with no/minimal human intervention. Semantic descriptions provide an opportunity to derive additional functionality simply by composing existing services together. Various composition approaches have been proposed for service composition. However existing approaches have failed to take the important aspects of preconditions and effects into consideration. This reduces the accuracy of the composition process and leads to false positives which must be avoided. In this paper we first motivate the need for precondition and effect matching in composition and identify the limitations of existing approaches. We propose an algorithm for composition which addresses problems in current approaches and employs the complete semantic description of inputs, outputs, preconditions, effects. We also propose a distributed architecture for implementation of the composer. Finally we demonstrate effectiveness of our approach by comparing with existing approaches.