The Research And Implementation Of Service Domain-based Automatic Service Composition

With the growing maturity of Web Service's technology, more and more services have been developed in the Internet. In the informative, complex and volatile Internet environment, how to manage enormous web services to quickly find suitable services, and how to make composited services adapt to service execution failure case, which caused by network fault or other failures, is becoming current research focus.In order to solve two problems described above, this paper introduces the concept of service domain to divide web service type into different domain, according to web service features. And then automatically composite web service domain or type to satisfy customers'requirement based on Hierarchical Task Network (HTN), which is Artificial Intelligence (AI) planning. Finally the composition results are executed by workflow to improve results'reusability.To achieve these research objectives, our research and related works are as follows:1. This paper systematically and comprehensively analyzes the existing Web Service description standards and Web Service composition methods, and compares the two main composition methods AI Planning and workflow with each other in detail. Meanwhile, we describe characteristic and disadvantages of AI planning composition based on service type.2. With the increasing number of Web Services which provide the same function, this paper propose a ontology description about service domain based on Ontology Web Language for Web Service(OWL-S). We construct an adaptive Web service domain model for matching interpretability characteristic of Ontology Web Language (OWL). With the growing number of service instances, the model can automatically adjust the classification structure of service domain. Experiments show that this model can improve the accuracy of service discovery, and enhance the degree of service management automation.3. Based on the service domain model, this paper provides a self-adaptive dynamic composition method which combines the AI Planning and workflow. This paper designs a set of rules, which map web service domain to description of HTN planner, to get feature supplied by combined service through HTN planning. Meanwhile, an exception handling module is added to achieve dynamic service replacement at web service runtime .4. On the basis of research above, a prototype middleware system is designed and implemented. Furthermore, a demo, Online Film Order, is developed based on the middleware. And we verify the feasibility of the Service domain-based automatic service composition through the demo.