Web service composition enables the creation of new and more valuable services to combine and link existing services. However, the treatment of user constraints (as user demands) is important to provide users with a more intelligent service. It is difficult for existing services to fully satisfy users' constraints, as user demands change dynamically, depending on user preference or environment.

We have produced an intelligent infrastructure that can deal with dynamic user constraints for Web service planning and composition. To realize this intelligent system, we analyzed the pattern of exceptions to manage the user constraints that can occur during planning through the creation of a scenario, and then we extracted important ideas for the detection and improvement of the exceptions. We also considered the system architecture using exceptional patterns and situation calculus theory in the first-order logic language to deduce the planning procedure. Our simple prototype used IndiGolog, which is a high-level programming language for robot, and applied situation calculus theory. Using this intelligent infrastructure enabled us to easily design a complex plan.