Web Service Selection Method Oriented By Domain Qos Constraints

There is a large quantity of services providing same functions in Internet, how to select one service from them to satisfy user’s requirement mostly is one of the key issues in this research areas at present. Most researchers think that when service consumer faces the problem of service selection, the system would consider not only functional needs but also non-functional needs, quality of service. The industry and academic do a lot of research work on QoS driven service selection separately from different views, and make a large number of achievements. However, the recent work on QoS still has some deficiencies. Firstly, the evaluation of QoS is focus on choosing attributes and evaluation algorithm, and neglects the influence by different steps in lifecycle of service. Secondly, there is less research on reliability of QoS information and its evaluation method. Thirdly, the present selection method divides the selection scenes into two categories, which are considering users’ requirement of QoS and not considering, and doesn’t take their relationship into consideration, then neglects the users’behavior for taking part in service selection. Fourthly, it considers the evaluation of common attributes QoS, and neglects the domain-oriented QoS in the process of QoS evaluation. Fifthly, when users provide the quality requirement in service selection, anyone focuses on the adaptive mechanism of service selection based on constraint.As a consequent, this dissertation provides domain-oriented extended framework for QoS, a multi-dimension evaluation technology of QoS and the extension to traditional SOA. Considering the uncertainty of services, it puts forward reliability evaluation method for supporting domain-oriented QoS with the information in different service lifecycle. To solve the problem of choosing abtributes of QoS, it proposes domain-oriented service constraint generation method based on users’using information. Finally, it provides an QoS driven adaptive selection method based on constraint relaxing model for considering users’constraint. The main contributes of this dissertation is summed as followed.To solve the problems on the existing QoS model applying in different areas, this dissertation provides domain-oriented multidimensional extended QoS model to support the increase, acquisition, transfer and storage of QoS information, and provides the foundation of domain oriented service evaluation. It also provides an extended architecture of SOA, which supports the transferring and evaluation of QoS information and QoS driven service selection.As the environment of service have the characteristics of openness and uncertainty, service provider would provide false QoS information. Considering the above problem, this dissertation provides the domain-oriented reliability model by calculating the differences between different QoS values in different steps in service lifecycle. The model takes the QoS and its influence in different steps in service lifecycle into consideration, and supports the reliability computation of services in multi-domains.This dissertation divides QoS driven service selection into two categories, which are considering users’constraint and not considering. According to the service selection having users’constraint, this dissertation proposes QoS model supporting domain characters, and select the best service to satisfy users’constraint mostly. It analyzes users’behavior on selection and feedback in selection process having constraint, and applies the analysis results into the selection process not having user’s constraint. This dissertation combines two categories with together, and solves the problem on choosing attributes and defining weight in QoS evaluation.Considering the problem on all of the candidate services not satisfying users’constraint and fails, this dissertation puts forward an adaptive QoS driven service selection method. It analyzes users personalized constraint, and provides a constraint relaxing model to support adaptive selection in service selection having users’constraint. The model analyzes the content and weight of constraint, and defines relaxing degree, equivalence and termination condition for constraint. The method could promote the success of service selection, and decreases the time for users’joining in it.