Identifying Optimal Composite Services by Decomposing the Service Composition Problem

For a Web service composition to satisfy a user's needs, it must not only provide the desired functionality, but also have nonfunctional properties (e.g., reliability, availability, cost) that are acceptable to the user. In the recent past, several techniques have been developed and deployed to identify a composite service that conforms to the functional requirements and is also optimal with respect to the user-defined preferences over non-functional properties. However, these composition techniques are limited to using one formalism for specifying the required functionality, in short, the existing techniques cannot identify optimal (w.r.t. non-functional properties) composite services that are required to satisfy functional requirements described in multiple formalisms. We have previously proposed a meta-framework for service composition that involves decomposing the required functionality into a boolean combination of atomic requirements, which are expressed using different formalisms. This meta-framework supports the use of multiple formalisms and their corresponding composition algorithms within a single scenario. In this paper, we integrate support for unconditional preferences over nonfunctional requirements into this composition meta-framework. We show that for a large class of problems, local selection of preferred service(s) can yield the most preferred composite service that satisfies the desired functional requirements.