Agent-Based MADM Approach to the Dynamic Web Service Selection

How to dynamically select and bind Web services to ideally meet the requirement of service consumers becomes a more and more important topic in Web service community. This paper presents an agent-based multi-attributes decision making (MADM) approach for the dynamic Web service selection. We summarize the Web service selection problem to MADM problem, and take an agent which executes periodically measurement tasks to avoid the imprecise of decision matrix (DM) as well as to remove unavailable services from the service registry. In addition, we introduce a simplified but effective method to work out the MADM problem. Experiments shows this approach is effective in particular for services which have multiple duplicates and require mass computation under unstable network environment.