Web Services Composition And Recommendation Leveraging Semantic Similarity Computing

With the wide application of Web services,the number of Web services is increasing,and a large number of applications are packaged in the form of Web services and can be invoked through the network.It has become increasingly difficult for a Web service with a single function to meet the needs of users.Therefore,it is a problem that needs to be solved by composing a service that conforms to a user’s needs through a service composition method.It usually takes a lot of time to combine the corresponding services.The Web service recommendation system can help users quickly find the Web services they need to combine and recommend.Based on the semantic similarity,this thesis explores and discusses the main issues facing Web services composition and recommendation.First,by changing the existing services and operations into a hierarchical structure through a defined method,using the semantic similarity method,the possibility of invoking between two operations in a service is calculated and the semantic similarity is achieved.Evaluate how the two operations relate to each other.Second,we built a Web service network model.The vertices in the model are operations in a Web service,where the direction of directed edges represents the relationship of invocations between two operations.The thickness of the directed edge indicates the possibility of calling between operations.The weight on the edge indicates the possibility of calling each other between two operations.At the same time,based on the Web service network model and the user’s functional description of service operations,a semantic similarity model is adopted to identify the user needs.Finally,a service composition and recommendation algorithm based on dynamic programming and a genetic algorithm based service composition and recommendationalgorithm are proposed.In the process of service composition and recommendation,the dynamic programming algorithm is decomposed into several sub-problems by solving the problem,and the sub-stages are solved in sequence to reduce the amount of calculation.In genetic algorithms,a variable-length operation has been added to allow the length of the service composition segment to change.Real number encoding and corresponding crossover operations are used.The accuracy of the genetic algorithm is improved,the speed of convergence is accelerated,and the time consumed is reduced.This thesis adopts a semantic similarity-based algorithm to establish a Web service network model.It combines and recommends services through dynamic programming and genetic algorithms.It evaluates through experiments and verifies the accuracy of service composition and recommendation algorithm.