Semantic Web Service Composition And Related Research Based On Ontology And Multi-agent Technologies

Service-oriented computing and service-oriented architecture are revolutionizing the software and network change. Web service composition is the core idea of SOC and the SOA; it is also the critical factor to promote SOA from concept to application. With the expansion of Web services by Agent, Agent and Web services can interact well to help Web services achieve high-level applications; meanwhile the semantic descriptions of Web service provide Agent with an automated serving and semantic reasoning method. In this dissertation, based on ontology, combined with Multi-agent technology, focused on Web service composition, the following research work has been completed.1. Task decomposition was implemented based on ontology and Agent technology.The deliberative Agent was studied and based on BDI structure an extended Agent model was proposed, which made Agent has the ability of representation the user behavior and preference. Combined with ontology and Agent technology, oriented task decomposition, a complex task can be decomposed into some simple tasks by using hierarchical analysis method, and then the Agent-based hierarchical task description language AHTDL was proposed. At last, with a practical example the proposed method can be analyzed and verified.2. The personalized Web service composition was generated based on ontology and Multi-agent technology.OWL-S was extended, and the formalization and quantifying formula were presented. Personalized user model was constructed. Travel ontology was constructed and the corresponding Jena rules were designed. On this basis, a personalized service composition framework was proposed, which can support automatic Web service composition and invoking and implement the personalized Web service composing and recommending.3. By using Multi-agent technology, the adaptability of service composition was improved.The Agent-based adaptive system architecture was proposed, in which agent can sense running context in real-time, fully understand the network environment, computing resources, service capacity and other factors. With the agent layer, complex network environment facing problems can be dealt with not much human intervention, at the same time the model was proactive and universal.4. The Multi-agent dynamic cooperation model was given in Web service composition.An extended BDI-based dynamic Agent cooperation model (BDACM) was proposed. An extended agent capability representation and evaluation were given, and by which the improved contract net algorithm was presented. KQML primitives were extended, which can satisfy high-level task cooperation and own more flexibility. An extended agent life cycle model was introduced to manage the agent life cycle, which improved system robustness.