Planning With Incomplete Knowledge for the Composition of Web Services

摘要

Web Services is an emerging paradigm in which very loosely coupled software components are published, located, and invoked on the Web as parts of distributed applications. Web Services provide a new way of distributed computing where the interoperability between diverse applications is achieved through platform and language independent interfaces. The main focus of Web Services is the ability to easily combine existing components to create compositions that provide novel functionality that was not directly available from the existing services. Web Services composition is useful for a wide range of audience: ordinary users doing everyday tasks on the Web, commercial organizations involved in e-business applications, and researchers doing intense scientific computation over distributed networks. Automated composition of Web Services requires fairly rich machine-understandable descriptions of services that can be shared between heterogeneous agents. Given appropriate descriptions, AI planning techniques can be employed to automate the composition of Web Services described this way. The purpose of this research is to show that AI planning techniques can be extended to automatically generate useful and purposeful compositions of Web Services under incomplete information. We have worked on how Web Ontology Language (WOL) can be used to describe Web Services. We have also studied how planning techniques can be used for automated composition of Web Services. In both systems, reasoning with Web Ontologies has been used to facilitate the composition task. Our goal is to investigate how these approaches can be combined with each other and can be extended to efficiently address the issues related to the nature of the Web.