Structure Pattern Oriented Method for Service Composition

Improving dynamic feature of service composition is a vital and challenging stage in web service application. With the increasing of service number, it is hard to balance the flexibility and verification very well in dynamic and complex environment. The previous methods compare two services by their own features. These methods are useful and concise to find substitutes for unavailable services, but they might be too strict to find a solution on some occasions. We propose a novel method called SPACE to build and estimate service composition. The method uses six composite patterns as basic elements to organize services and it also measures impact of service replacement on these patterns in stead of pair of single ones. By using SPACE, we can find non-optimal solution for service selection and re-planning in service composition.