Composite Service Stage Optimization System Based QoS-Rules Research And Implementation

Web services with a strong autonomous, loosely coupled, coarse-grained, openness and integration features. Web services are reusable software modules, only need simple coordination, allow free configuration. At the same time, All the public agreements of Web services fully use open standards to describe and transfer. It can mask differences between different platforms. With the increase of available services, more and more requests need to be completed by a combination of services. There were many different service providers provide the same functionality of the service, but different performance services. Assessing the quality of many services is a difficult problem.In response to this problem, this paper uses data mining techniques search for the law according to the past running information, and also predicts the performance of Web services according to these laws, and recommends suitable services based on the predictable performance for system, and generates the instance of the initial implementation. This paper raises a QoS rule antecedent which based on service operation parameters and network status parameters, and also raises a QoS rule conclusion which assess the performance of QoS parameters. Such a rule vector expresses the QoS performance of service in the current state very well. In the final, these methods in this article have been realized, and are also used in the prototype system.This system implements the effective integration between service execution logs and network status monitor logs, and makes the original data set with a single logical. The system deletes the attribute which is unrelated with the QoS parameters, to reduce the mining operation. Second, to discrete condition attribute with continuous data types, and divide the value of the property into intervals, and use the label to replace the range of values, and make mining results simple, easy to use. Third, the system use the FP-tree algorithm which not generate candidate frequent item set to mine the data set, and also generate the QoS association rules, to find the link of condition attributes and QoS attributes.In the final, this system build the classifier based on QoS association rules, generate a decision table and predict the performance of Web services, and also generate the initial implementation instance.