Research On Ontology-Enabled Web Services Intelligent Negotiation And Monitoring Mechanism

Web service negotiation is an important stage in service interaction. At this stage, service providers and consumers reach an agreement on one or more service related issues and delivery details by continuous communication and decision-making, and then sign a formal service contract. In recent years, along with the technology development of Web services, computer science, artificial intelligence, and semantic Web, Web services negotiation is getting attention and becoming more and more automatic, intelligent and semantic. Along with the signing of service contract, if the factual qualities provided by service provider are in accordance with the contract become the main concern of service consumers, therefore, a Web service QoS(Quality of Service) monitoring tool is needed, which can constantly monitor and collect the factual quality data of the service.Based on the background mentioned above, our research focuses on ontology-enabled Web services intelligent negotiation and monitoring mechanism. In this paper, we take the negotiation ontology, negotiation strategy, service monitoring and evaluation as the subjects. The main works and contributions of this thesis are as follows,(1) By applying the ontology technology in the Web services intelligent negotiation, this thesis constructs an Ontology-Enabled Web Service Intelligent Negotiation Framework. We describe the components and functions of the negotiating framework and give the formal definition of the negotiation Agent. Based on this framework, the whole negotiation process is also illustrated in this thesis.(2) This thesis completes the definition and reasoning on negotiation issues ontology, negotiation protocol ontology and negotiation rules. We extract the class and properties hierarchies on Web services negotiation issues and describe it by using OWL (Web Ontology Language), then the Pellet inference engine are used to complete consistency checking on negotiation issues ontology. This thesis also defines the Web services negotiation protocol ontology related concepts and rules based on the OWL and SWRL (Semantic Web Rule Language). We combine the SWRL rules with JESS (Java Expert System Shell) reasoning engine to implement the negotiation rule reasoning. These researches lay the foundation for the Web services intelligent negotiation system.(3) A negotiation opponent behavior assessment and prediction algorithm is proposed. We build multi-attribute utility function and determine the weights in the utility function by using analytic hierarchy process (AHP), then the multi-attribute utility function is used to implement the negotiation propose evaluate, which guide the decision of negotiation Agent. This thesis also calculates the negotiation opponent’s approximate propose function based on the opponent’s historical proposal by using the polynomial interpolation method, so that we can predict the next proposal of the opponent. Through the experiment, we prove that this method is feasible and effective, which explores a new idea for the current negotiation strategies research.(4) To provide an automatic approach to monitor and collect QoS values of the service, an independent Web services QoS Monitoring (QoS-M) tool is developed. The tool can automatically parse the WSDL (Web services Description Language) document, generate the service invocation data and client stub code, and implement the automatic call to the Web service. Monitoring aspect codes are weaved into the Web service invocation point, which can dynamic collect the Web service call time and call state, then compute and storage the QoS dynamic parameters values. QoS dynamic updating algorithm is also used to calculate and update historical data information periodically, which reflect the latest changes of QoS information.(5) This thesis compares the quality data provided by service provider with monitored by QoS-M and implement an algorithm to compute the similarity, and then the reputation levels of the Web service are given based on the similarity.