Research On Semantic Sparse Web Service Discovery Technology

With the popularity of Web service applications,the number of Web services has increased exponentially,and Web services,as a reusable and loosely coupled distributed computing model,have been widely concerned.The continuous emergence of service market has further promoted the utilization and development of service resources.How to quickly and accurately find out the Web services needed by tens of thousands of Web services has become a hot topic in academia and even in industry.Web service often exists the problem of sparse representation of service description and vague expression.This phenomenon of semantic sparsity can lead to ineffective computation of similarity.At present,there are four main methods of Web service discovery:keyword matching,semantic matching,behavior matching and clustering based matching.In these four methods,few researchers focus on the research of Web service discovery method for semantic sparsity.Aiming at the problem of sparse semantics in Web services,two models are proposed.One is the Web service discovery model based on word2vec semantic expansion and LDA text clustering.The training set is trained by word2vec,the word vector is calculated,and the text is expanded by the similarity between words and words to enhance the text's richness.Then,the extended text is clustered through the LDA theme model,and the similarity degree is compared between the subject words and the retrieval formula,and the similarity is the largest.Class clusters are displayed to users,which improves the accuracy of retrieval.The other is a service discovery method based on the feedback recurrent neural network and the word embedded topic model.This method first generates a word embedding model by using the feedback neural network,and then designs a word embedded topic model that combines the label knowledge to provide support for the service clustering organization;finally,according to the service hierarchy model and the service hierarchy model,User query requirements,a probabilistic query strategy is proposed to achieve efficient service discoveryUsing a set of data sets to verify the effectiveness of the algorithm,the data set uses a web crawler to crawl the real Web service data registered in PWeb,and designs two groups of contrast experiments on this set of data sets to verify the effectiveness of the two algorithms.In the experiment,two Web service discovery models are compared with the traditional Web service discovery method based on keyword search.The experimental results show the effectiveness of the proposed algorithm,and it is proved that the semantic expansion of Web service description can improve the accuracy of the algorithm.