Abstract: With the rapid growth of functionally similar Web services over the Web,Quality of Services(QoS) is becoming a decisive factor for Web service selection. However, current QoS measurement method generally takes the mean value of Web service history QoS values as the service's QoS, and hasn't taken the dynamic nature of service performance into consideration, and cannot measure QoS of Web services accurately, as a result, many selected services cannot satisfy consumer' s QoS constraints. In our framework, a dynamic QoS prediction method based on Cascl3ased Reasoning(CI3R) was provided, this method associates Web service QoS with the environment of service, task type and task size together, when a new service request comes, CI3R is applied to predict QoS of the Web service in completing the new request Experimental results show that, this method can improve accuracics of QoS of Web service effectively.

Key words: Web service selection, QoS dynamic prediction, Case based reasoning