Research On Web Service Discovery Based On Knowledge Map And Word Embedding

Web services,as a reusable,loosely coupled distributed computing model,have received extensive attention from academia and industry in recent years.There are some problems such as sparse service description and fuzzy semantic expression of Web service.Because of sparseness of Web service description,the user cannot accurately and efficiently find out the Web service meeting the requirement of personalization.At present,there is little research on web service discovery in the background of semantic sparseness.In order to solve the semantic sparseness problem in Web service,this paper proposes the basis.Web service discovery methods embedded in knowledge maps and words.The main results are as follows:(1)one is a topic distribution model based on the extension of synonyms in knowledge graph.Firstly,the preprocessing of Web service features is carried out,including word segmentation of service name,service label,service description,part of speech reduction and filtering deactivation words,and then the preprocessed Web service document is extended by knowledge graph wordNet to enrich the semantics of service description.Finally,the extended text is clustered by LDA clustering algorithm,and the service is organized to different potential topics.According to the similarity between the retrieval word vector and the subject word vector,the Web service of the same topic is selected and recommended.And the searching rate and the full rate of the Web service query are improved.In order to verify the validity of the model,several groups of comparative experiments were designed by using the real-registered Web service data in the public test data set(OWLS-TC).The experimental results show that the semantic-extended Web service semantics are richer and the description is more specific.Compared with the traditional keyword query method,the calibration rate is improved by about 10%,and the experimental results show the effectiveness of the proposed algorithm.(2)the other is a word embedding topic model based on the semantic similarity,and first,the training model word2vec is used for carrying out word vector training on the characteristic words of the service description to obtain a continuous word vector set;then,the semantic expansion is carried out on the text by the similarity between the calculation word vectors,and the semantic of the expanded Web service document is richer;and then the characteristic word is mapped into a word embedding by traversing each characteristic word of the user query,Using the Gaussian LDA clustering algorithm,the service organizations of the same subject are clustered together,and the efficiency of the Web service query is more efficient.real The test results show that the precision and recall of word embedding topic model based on semantic similarity are improved to a certain extent.Compared with the extension of synonyms in knowledge graph,word2vec semantic expansion effect is more ideal and query efficiency is higher.