| Web Services provider a chance for system application in achieving cross-platform alternation and integation, and become the focus of industry. Yet the function provided by single Web Services is limited, it is necessary to composite Web Services to provide more functions. In Web Services Composition, it is possible that different providers deliver similar or the same services due to the increasement of Web Services providers. In order to assist user in Web Services auto-selection and composition, QoS is taken as the criteria of Web Services differentiation and evaluation on the basis of fulfill user requirement. In fact, QoS has several properties and it is complicate to compute. The key problem is how to combine QoS with proper algorithm in order to generate the services which could meet user requirement. Meanwhile QoS weaken by non-functional services invalidation in services dynamic change also brings uncertainty in feedback of successful services plan. This thesis is mainly work on this problem, and its main content including:Firstly, it introduces a QoS driven Web Services composition system framework. Taking five typical properties as example and considering plan related thinking, it takes plan selection algorithm in and composes the composition system. Services composition is divided into three main parts including: initialization, composition and multi-services disposition. It combines QoS selection logic with algorithm and uses QoS to direct the whole process of composition.Secondly, under this framework, this thesis presents multi-properties power additive comprehensive composition and priority-differentiation composition. The prior transits porperties in [0, 1] and then add power to compare. The latter focuses on users' perference on porperties and their acceptance on hypo-plan. Comoposition describes logic through specific algorithm, and uses examples to illustrate the work flow. It can return user satisfying plan in nice time complexity.Thirdly, in order to solve the non-functional services inefficient problems caused by services dynamic changing, it introduces a set of services properties combination mechanism depending on application environment after considering the internal relation of services properties. Using caching plan, this mechanism balances the property influence on final QoS and fulfill the user requirement. |
