Research On Web Service Composition Algorithm Based On QoS

Web service as a distributed application technology attracts more and more attention. On the one hand, with the constant changes of user requirements, atomic service which provides a single function cannot meet the requirements, only in the way of combining of existing atomic services, can it provides value-added services to meet user requirements. According to the functional requirement, Web service composition combines the existing atomic services to provide comprehensive and complex services. On the other hand, with the rapid development of e-commerce, enterprises pay more and more attention to user experience about quality of service(Qo S). Therefore, on the basis of meeting user functional requirements, how to choose service instances reasonably and efficiently to compose composition services with the high overall Qo S performance is a problem to be solved.The premise of Web service composition is to select the Web services which match the user functional requirements among a large number of atomic services. The traditional Web service matching based on the keywords of services of request and candidate. Because of the lack of semantic, the keywords based on syntax cannot be understood by machine, which causes a very low accuracy.Given the existing semantic Web service matching is not good enough, which can barely match correct words, this paper puts forward a new kind of semantic Web service matching algorithm based on hybrid strategy. The algorithm combines a new semantic distance similarity algorithm(based on the logical match), the semantic matching degree coefficient overlapping degree algorithm(based on the illogical match) and a particular service processing without input or output string similarity algorithm named Levenshtein Edit Distance Algorithm. Experimental results show that after the combination of the three strategies, hybrid matching algorithm improves the precision and recall ratio.Multi-objective Ant Colony Optimization for the basic MOACO in solving the the Web service selection problem based on Qo S is easy to fall into problem of local optimum, in addition, the search efficiency is not high and it does not take the disadvantage of user preferences into consideration, this paper presents a new Multi-objective Ant ColonyOptimization for N-MOACO. Specifically, first, the algorithm will make Q pheromone intensity self-adaptive, solve the problem of local optimal; Second, the algorithm will reduce the number of candidate services, so as to reduce the number of unnecessary search when the ants choose the next service by using the correlation between Qo S properties according to the reasonable Qo S attributes range of candidate services of the current service Qo S prediction; Third, the algorithm puts forward a new heuristic function, the heuristic function is related to the Qo S attributes for price, the excution time and reliability of the Web service, the goal of optimization is to get a better user experience. At the end of the paper, a scheme of the highest customer satisfaction can be made from combination plans of N-MOACO algorithm by using the AHP analytic hierarchy process. Experimental results show that the solutions required by N-MOACO algorithm excels solutions required by MOACO in terms of dominance and user satisfaction, and the search efficiency of N-MOACO algorithm is higher than that of MOACO algorithm.