Study On Key Techniques For Dynamic Web Services Selection Supporting Relation QoS

Web service is an application program, which is based on internet, distributed, self-included, self-described and modular. It possesses the attributes of well encapsulation and loose coupling. With the development of Web service, single service from suppliers is becoming hard to meet the demands of the users. So Composite Web Service which can provide value-added functions has received a lot of attentions. Service composition emerges as the times require. Service composition is to combine the relative service dynamically in term of situation requires, and come into being a new service to meet the demands of the users.However, with the prevailing of Web service, the number of services on the internet has been larger and larger. How to select a set of services among large body of services with the same function in order to make the Composite Web Service achieve better performance and better quality is a hot issue in recent research.The thesis mainly discusses the dynamic Composite Web Services selection, and studies dynamic Web services selection approach supporting relation Quality of Service (QoS) in the semi-dynamic composition situation.Firstly, in the study of QoS the thesis analyzes current QoS models' advantages and disadvantages, based on which proposes a new, objective, reliable QoS model called an Extended QoS model (E-QoS model). The E-QoS model presents the relation QoS between services which is not only from the view of composition efficiency, but from the relationship of the services to study the terms to be satisfied. Relation QoS effectively ensures the Composite Web service to implement correctly. The E-QoS model regards the relation QoS between services, the domain level QoS and the service level QoS as important QoS factors to value quality of Composite Web Services. And the model plays a foundation role in selecting the most qualified Composite Web Services.Secondly, the thesis presents a dynamic Web services selection approach supporting relation QoS in the semi-dynamic composition situation. The approach is based on genetic algorithm and uses the E-QoS model to solve the composite optimization, which guarantees not only holistic optimum, but also local optimum. In order to improve the efficiency of the approach, the paper uses Web service class conception and service selection based on ontology to select candidate services. Statistical experiments prove the optimized approach possesses the attributes of better validity and better stability.Finally, the approach is used in Smartflow pattern and shows good performance.