A planning approach for message-oriented semantic web service composition

In this paper, we consider the problem of composing a set of web services, where the requirements are specified in terms of the input and output messages of the composite workflow. We propose a semantic model of messages using RDF graphs that encode OWL ABox assertions. We also propose a model of web service operations where the input message requirements and output message characteristics are modeled using RDF graph patterns. We formulate the message-oriented semantic web service composition problem and show how it can be translated into a planning problem. There are, however, significant challenges in scalably doing planning in this domain, especially since DL reasoning may be performed to check if an operation can be given a certain input message. We propose a two-phase planning algorithm that incorporates DLP reasoning and evaluate the performance of this planning algorithm. Copyright © 2007, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.