Semantic Web Service Selection Based On Improved Fuzzy Algorithm Supporting QoS

With the rapid development of Web Services technology in Internet, the numbers of Web Services which have the same function are shutly with the day. Users not only take care of the function but also the non-functional factors of services, especially the QoS (Quality of Service), when select the services. However, there are many drawbacks in traditional keyword-based selection methods such as low recall, low precision and so on. The semantic is introduced into Web selection mechanism. By using ontology technology to describe the QoS information, the capability of semantic description QoS has been enhanced. And the matching degree between the required services and candidate services has also been inproved.The semantic and numerical characteristics of QoS attributes are studied in this paper. The OWL-QoS ontology, which has good expansibility, is constructed at the same time.This ontology contains single-valued, fuzzy single-valued and interval-type measure methods. It also supports numeric and language-based metric values. The constructed OWL-QoS ontology is used for semantic match of QoS parameters which from service providers and service requestors. According to the parameter values of service requestors, the evaluation matrix will be made after choose proper candidate Web Service and filter the experts'ideas.Related theory and methods in Fuzzy Mathematics are studied. The weight value is counted by using Analytic Hierarchy Process. The improved fuzzy membership evaluation algorithm based on membership conversion is proposed from the perspective of redundancy, achieving membership conversion with no redundant data. The quantitative and the qualitative factors in the index system are combined. It has enriched the original evaluation factors of QoS.The improved fuzzy comprehensive evaluation evaluates the candidate service in QoS attributes factors, whice based on membership transformation. The rate of real-time load affects the executive efficiency of service. It is analysis by synthesis in this paper. The comprehensive values of OoS evaluation value and the real-time load rate are counted. The final result in descending order is returned to the service requestors. Finally, extensive experiments show the superiority of the proposed method.