Research Of WEB Service Quality Evaluation

Quality of Service (QoS) is an important factor during service composition and recommendation. However, at current stage, most of the researches of QoS focus on the static quality of services and the description of QoS through expanding WSDL. Little attention was paid in the fields of how to obtain the dynamic quality of services and the metrics of QoS, which are also obviously important in the expansion of services and the extension of service quality attributes. Since static QoS data cannot reflect the real-time performance of Web Service (WS), a mechanism is needed to collect the dynamic QoS data. Also, QoS query result is determined by the QoS data stored by service registry. Consequently, it is also necessary to use an efficient and accurate feedback method to calculate and send those dynamic QoS data to the service registry.This paper is supported by 863 program under Grant No. 2007AA01Z139, Research on service description, selection and evaluation based on closed-loop feedback mechanism under Internet environment. In this paper, we propose three methods to collect the real-time QoS data, namely AOP-based, Proxy-based, and Port-based. The application of the three methods will be discussed by comparing their advantages and disadvantages. In the implementation of specific application, we discussed the advantages and disadvantages of these methods. These methods have their own application environment, through their integrated applications, allows us to greatly enhance the adaptability of feedback. Since the feedback conforms to the OWL language, which is a rich and extensible modular ontology language, programs can understand the semantic feedback automatically by parsing the feedback messages.And we propose a QoS feedback model based on objective QoS metrics using some simple statistical theories and a dynamic queue as a data pool to cache all runtime status. Moreover, error determination and sampling feedback have been taken into consideration so that service provider assigns less hardware resource and avoids disturbing feedback result from unfriendly exception. By carrying out experiments, it demonstrates that this feedback model evaluates the WS performance better than other common methods. This model provides QoS metrics that are easy to rank and sensitive to the status change.Finally, we introduce software based on our theory. We call it Q-Spy. Implementation of the tool increases collaboration between client and the service registry through feedback. Add this collaboration because we found that in most cases, the client feedback on the quality of service through customer similarity analysis method, is much better than the service provider feedback data, for consumers to find more suitable services through the quality of service.