Skip to Main Content
A novel two staged and user requirements oriented Dynamic Web Service composition framework is presented in this paper. Firstly, we separate the end user requirements into two parts:functional and non-functional requirements. Consequently, with user requirements in OWL-S, the composition firstly uses JSHOP2 technique to generate an Abstract Service Plan to satisfy user functional requirement. In the second stage, the Web Service Instance Selection problem is converted into a multi-objective optimization problem, using the Multi-Objective Ant Colony Optimization (MOACO), the Abstract Service Plan is concretized into an Concrete Workflow based on non-functional requirements, and then being transformed into BPEL4WS. Theoretical analysis indicates the proposed solution which separately dealing with the functional requirements and the non-functional requirements leads to a significant reduction of the search space. The prototype in a domain-specific scenario indicates the feasibility of the framework.