Research On Efficient And Trusted Dynamic Web Services Optimization Composition

With the development of cloud computing,Internet of Things and other technologies,more and more Web services are shared on the Internet.However,a single service is often unable to meet the complex business requirements of users.It is an effective way to satisfy the requirements by aggregating the available services into more powerful composite services.The openness network environment shared by Web services and the competition of service providers result in the instability and unreliability of Web services.Therefore,how to select the optimal trusted services to meet the user's requirements from the large number of candidate services has become a key technical problem to be solved in the field of Web services composition.In order to improve the efficiency of service selection in the case of a dramatic increase in the number of services,and to ensure the credibility of the selected services,combining the dynamic nature of the service composition and the relationship between candidate services,a two-stage service composition method of ‘ Efficient and Trusted '-‘ Dynamic Correlation ' is proposed.This method integrates domain template technology,trusted service recommendation technology,dynamic Qo S constraint decomposition technology,and service correlation aware technology.First,the domain template technology is used to perform candidate service filtering,and the candidate service set is reduced.Second,social network trust model and collaborative filtering etc.recommendation technologies are used to recommend trusted Web services from the reduced candidate services set.Finally,combining the user demand and global Qo S constraints to decompose the evaluation model,the global Qo S constraint is dynamically decomposed into local Qo S constraints,and service dynamic optimization combination is performed based on the relationship between constraints and candidate services.The research work of this paper mainly includes the following aspects:a)In order to solve the problem of low efficiency problem of the traditional composition method while composing small granularity,general and dispersed service components with the sharp increase of services number,a service filtering method based on domain template is proposed.First,combining the domain knowledge of service industry and business background,analyzing and mining the domain characteristic of services,and abstracting,summarizing,classifying,and summarizing the common demands of services field,a domain template based on swarm intelligence computing technology is constructed.Second,the Web service quality evaluation model including service domain attributes is designed to evaluate and select the template.At the same time,the ant colony algorithm is improved to ensure the timeliness of the domain template.The pheromone enhancement method and ant colony mutation learning function are added to make up for the deficiency in convergence and local optimization,and to realize the domain template updating automatically.A service filtering method based on domain template is proposed by using the matching rules.After the candidate service set is filtered by the domain template filtering technique,its space is reduced for judging the services credibility.b)In order to guarantee the credibility of composite services provided to users,the paper proposes a method of recommending trusted Web services under the social network environment,based on the techniques of social network and matrix probability decomposition.First,according to the information of user evaluation deviation,confidence interval,malicious degree and comprehensive malicious degree,the malicious evaluation information is eliminated to ensure that the evaluation information is not disturbed by noise.Second,based on the trust relationship in social network,a trust relationship model is established,and the subjective trust and objective evaluation information of users are integrated.The degree of users' trust similarity is measured by the similarity and trust degree to improve the ability of the system to identify the nearest neighbor users,avoid the dissimilarity between the nearest neighbors caused by sparse evaluation information,and improve the credibility and accuracy of recommendations.Third,the cold start problem of the trusted service recommendation is solved by bring social trust relationship and rating information into the probability matrix decomposition model based on the collaborative filtering.Trusted service recommendation provides a set of trusted candidate services for dynamic Web services optimization composition.c)In this paper,we propose a dynamic optimal service composition method which is based on global Qo S constraint decomposition and correlation awareness.First,the global Qo S constraint is dynamically decomposed into local Qo S constraints by considering user requirements and quality scale evaluation model of global Qo S constraint decomposition and using the improved artificial bee colony algorithm so that it can be applied to the global Qo S constraint decomposition when there exists association relationship between candidate services.The artificial bee colony algorithm is improved by changing the initial food source and adding the search function with the service association data set in the employment bee phase and the detection bee phase.Then,the service association data set is constructed,and Qo S correlation coefficient between the upstream and downstream services is determined.After the Qo S constraint values of the candidate services are calculated the candidate services are selected which can meet user's requirements by association relationship between selected services and candidate services and the actual value of Qo S constraints.