Research Of Matching Mehthod Based On Monitored QoS In Semantic Web Services

With the increasing of the number of web services,how to distinguish these semantic web services with same or similar function effectively becomes an urgent problem to solve. In recent years, QoS(Quality of Service) is used as a standard to evaluate and measure the service. In general, the service QoS parameters are provided by user or service provider, but their credibility is not high, summarized in the following reasons. First, from an objective side, because of variability of the Internet Environment and operating environment, and constraints of existing web service implementation, and the other factors, the actual web service call may not provide its statement QoS. From the subjective side, in order to attract more consumers, the service provider may deliberately exaggerate QoS; for users, it hasn't yet to form a sound and effective incentive mechanism to encourage users to provide true and reliable information on QoS parameters, and if on their own perspective, they may provide false information. How to get true and reliable QoS parameters and select the most appropriate web service for the users has become the focus of academics attention in recent years.In order to obtain QoS parameters with high confidence, and return back the service which can meet user needs best, the thesis conducts a research on the following aspects:First, with the combination of web service monitoring and the non-functional matching needs, the thesis presents a QoS model with general and domain QoS parameters. Meanwhile, in order to make the computer understand QoS parameters with different formats provided by different users and providers, the thesis presents a new QoS ontology model with good semantic description and extensibility to regulate the QoS information description.Secondly, the thesis extends the semantic web service execution engine provided by OWL-S API through observer pattern in design patterns, so that in the course of execution it can send out some monitoring events on the key points, then specific QoS parameters which contained in the QoS model can be calculated with monitoring data. Meanwhile, the thesis also a specific QoS calculation method base on the data.Thirdly, because of the variability of the internet and other factors, the service returned through matching the monitoring data derived from any one of execution with user needs may not meet the actual needs of users. Therefore, in order to provide reliable QoS parameters, we use the IOWA operator to assess the historical monitoring data, then, match the predictive value with user QoS needs, thereby the accuracy of matching is improved greatly. Meanwhile, in the matching process, this thesis presents a matching method considering both semantic and value, thus it can be in line with actual QoS real reuqirments.At last, the thesis gives a detailed design about monitoring module in the monitoring system, and also the experiment is given out to verify the matching algorithm. The experimental results show that the algorithm is effective.