Research On The Approach Of Web Service Composition And Optimization Based On Logic Petri Net

With the maturity and popularization of the new generation technology including cloud computing,Internet of things,mobile Internet etc.,the development and application of software are with a prominently service-oriented feature.Web service has the important application prospect in service-oriented architecture,distributed computing,software and data reuse,and so on.With the increasing of number of Web service,how to combine the existing Web services to build new value-added services quickly,and meet the needs of users,become the research hot spot.There are lots of services with the feature of similarity on the function and parameter interface among the massive service repository.It can enlarge the granularity of service discovering,reduce the service searching space and improve the searching efficiency by mapping the similar services to a cluster.This dissertation proposed a novel model of Web service cluster using logic Petri nets as a formal tool,and studied the technology of Web service discovering and composition using this model.The main contributions and innovations of this dissertation are as follows.(1)Propose a similarity computation method based on tree structure coding of ontology library.Establishing an efficient domain ontology concept repository is an important basis for semantic-based service discovery.This dissertation proposes a similarity calculation method based on tree structure coding of ontology library.By encoding ontology concepts according to the hierarchical structure of ontology tree,the semantic distance of ontology concepts can be converted into character encoding distance,and the complexity of concept similarity calculation in the process of service clustering and service discovery can be greatly reduced without finding ontology tree.(2)Propose a novel model of Web service cluster based on logic Petri net.The definition of logic Petri net is improved to express logical expressions with logical vectors.It is convenient to improve the computational efficiency when judging the triggering conditions.Based on the theory of logic Petri net,the service cluster is regarded as the transition of logic Petri net,the service parameter set is expressed as a unified logical vector based on service cluster parameters,and the model of Web service cluster is established.In service matching,the set of parameters in the user's request can be transformed into a logical vector based on the parameters of the service cluster,and then the matching can be judged by logical comparison.In this way,semantic-based similarity matching can be limited to the level of service cluster comparison,and logical vector-based matching can be used within the service cluster,which can reduce the scale of parameter semantic matching and improve the efficiency of service discovery.(3)Propose a service composition architecture and discovery approach based on logic Petri net service cluster.A combination method of static and dynamic composition based on logic Petri net service cluster is proposed.Dynamic composition method is based on logic Petri net reachability calculation method.It searches service clusters and services in service clusters according to the idea of parallel layering,and establishes service composition planning.The static method is to save the successfully discovered service cluster composition structure and gradually establish the service cluster composition library.In the process of service discovery,the service cluster composition library is searched first.If the service cluster composition satisfies the user's request,the service cluster search service will be entered according to the service cluster composition structure,otherwise the dynamic search will be carried out.The combination of static and dynamic combinations can improve the efficiency of combinatorial discovery.(4)Propose an optimizing method of service composition based on both QoS and logic form transforming.If a planning graph is built in the process of service composition,optimal service composition can be got by reverse tailoring on the planning graph.In the reverse tailoring process,a threshold based on the best composition QoS can be dynamically determined,and can be used to filtering services to get optimal service composition with best QoS.The reverse tailoring process can be denoted as a process of transforming a conjunctive normal form to a disjunctive normal form.It can use logic fonnula to filter redundant service and ensure to obtain the most simplified composite service.Using service discovery and composition method based on logic Petri net service cluster and service composition optimization algorithm based on QoS and logical operation,service space and the scale of semantic comparison can be reduced,the computing speed can be improved,and the efficiency of discovery and composition of massive service databases can be improved.