The Study And Implementation Of Web Services Composition

As the development of the Internet technology, Web services provide an excellent technical support for sharing and processing information. Now, Web service has been attracted more attention duo to its superiority in standard interface, platform independence, standard protocols, etc.The function of a single Web service is simple. With the increasing demand for large and complex web services, the conception of service combination has emerged. It can not only make full use of the existing resources on Internet, but also improve the utilization of resource and reduce the time and cost. Therefore, the combination of service has become one of the main studying hotspots in the area of information sharing. The traditional service discovery technology uses keywords or the framework-based model to match and find required services. However, the precision and the recall ratio of them are low. Fortunately, the technique of semantic web makes the machine understand the information of service, which can improve the efficiency of the service matching to some extent. In the process, how to design service matching algorithm or model with more quick and accurate performance is one of improtant purposes of service composition.In order to solve this problem, a kind of service classification filter strategy is presented and an OWL-S/UDDI conversion module is constructed, which can add the description of the service'information, such as classification, function, and other semantic-based information.In the process of service matching, we summarize previous related research work, and adopt information matching and constructing the ontology tree to gain the service collection by multiple filter strategy. The result of previous filter is also the candidate set of the subsequent filtering process.Furthmore, a QoS based selection mechanism is proposed which can be used to filter the best services when there are multiple services with similar functionalities. By defining the concepts of user attention degree, quality factors and so on, a selection model and algorithm for filtering optimum services have been designed.Finally, the simulational experiments are designed to evaluate the method by using response time, recall and precision ratio as three key evaluation criterions. The experimental results show that the method is feasible and effective.