Research On Sevice Sunstiution Combining Service Cluster And Process Collaborstion Relationship

With the maturity of new generation communication technologies such as cloud computing,Internet of things and 5G,a large number of network services represented by web services have been published on various cloud platforms.The software development and deployment mode of service-oriented architecture(SOA)has been widely used.By invoking existing services,we can quickly build a new business system.For some complex service requests,we can build service processes through service composition to achieve demand response.However,the service is located in the dynamic network environment,some individual services in the working state may fail.At the same time,with the update of service version and the change of business requirements,it is required that the service substitution can be realized quickly in the cloud platform.Although the service substitution has achieved fruitful research results,the existing methods mainly focus on the service function and service quality.The search efficiency of the substitution service is not high,and the collaborative relationship between services is not considered.The rationality of service substitution needs to be further improved.In view of the above problems,this paper proposes a service substitution method based on service cluster fusion process collaboration.The main work and innovations of this paper are as follows:(1)A service collaboration sequence sampling method is proposed.The service calculation method for collaboration similarity is constructed based on Word2 Vec.The logical Petri net is used to model the service process as a service net.According to the different structures of the service net,a service collaboration sequence sampling method is established to ensure the complete sequence sampling of the collaboration relationship in the service process.The sampling sequence is converted into a corpus,and Word2 Vec is used to train vectors for each service node.The collaboration similarity between services is calculated through cosine similarity.(2)Service clustering is introduced into the service substitution process,and a service substitution algorithm integrating service function,quality and collaboration similarity is established.A service request response pattern oriented service clusters is established.The service clustering method based on LDA model and k-means++ algorithm is constructed.By the introduction of service clusters,the search speed of candidate substitution services is improved.A service substitution algorithm integrating service function,quality and collaboration similarity is established.The comparison experiments were carried out based on the mashup service data set and simulation data set.Experimental results show that this method improves the efficiency of service substitution and increases the rationality of service substitution.