Research On On-demand Services Discovery Approach Based On Domain Ontology And Spectral Clustering

With the development of software as a service and service-oriented architecture technology,the amount of available Web services resources on the Internet is increasing and widely used in software development.However,in the face of a large and growing number of Web services on the Internet,it is still a key problem to help users accurately and efficiently find the services they need.Studies have shown that some existing service discovery methods have not solved the problem of user query ambiguity,and lack of semantic extension to user's functional queries.In addition,the discovery of Web services is based on matching services in the whole corpus,without effective services classification and organization,resulting in low retrieval efficiency.In response to the above issues,this paper mainly carries out the following research:(1)Domain ontology is constructed based on association rules and improved K-means algorithm.Firstly,based on the domain vocabulary sorting table obtained by the iterative service classification using SVM,the top h words are selected as the domain concepts,then use the association rules and weights to construct the concept vector,and the improved K-means algorithm is used to cluster the concept vector to get the initial domain ontology;Lastly,the ontology is enriched by WordNet,which provides the foundation for the user's query expansion when the Web service is discovered;(2)Based on improved similarity computation,the spectrum clustering is used to complete the clustering of services in the domain and mining the theme of the service.The method is based on the principle of the similarity propagation of nodes in the network,we find a set of documents that are more similar to each document by setting the threshold,and further calculate the similarity between each two document sets by using Jaccard coefficients,according to the similarity matrix and the reference K partitioning problem in graph theory,we use NJW algorithm to cluster Web documents,and realize the topic extraction after clustering,which is to reduce the search space and improve the efficiency of service discovery.(3)Based on the above research contents,according to the strategy of “user query-ontology expansion-topic matching-service matching”,on-demand services discovery is conducted,and the feasibility and effectiveness of the approach are verified by the dataset on ProgrammableWeb,and the Web services retrieval system based on domain ontology is implemented using Java language.The Web service discovery method based on domain ontology and spectral clustering proposed in this paper can,to a certain extent,guide users t can to clear their needs and solve the problem of ambiguous user queries.At the same time,it can effectively organize services,reduce the search space,and improve the efficiency of Web services discovery.