Semantic Analysis Of Network Buzzword Based On Knowledge Graph

As a new and relatively self-contained special language,cyber language is favored by the vast number of netizens because of its colloquial,unrestrained,humorous and quick features.Because the knowledge graph has the advantages of simplicity,efficiency and speed when describing the relationships between words or sentences,the use of knowledge graph for semantic analysis can solve the problems such as the confusion of concept elaboration and the unclear relationship with common daily words in the process of popularization and use of network buzzword However,in the current process of knowledge graph construction,there are some problems,such as the low accuracy of named entity recognition algorithm for entity recognition of network terms,and the result extracted by relational extraction algorithm is not consistent with the fact.First of all,in view of the low accuracy of the commonly used named entity recognition algorithm for Network buzzword recognition,this paper identifies the entities of network buzzword based on the BILSTM +CRF model,which solves the problem that the sequence tags generated by the location tags of the Softmax layer do not conform to the grammar specification.In order to verify the reliability of the model,this paper uses the corpus of cyber language to carry out named entity identification extraction.The experimental results show that the BILSTM +CRF model can effectively improve the recognition accuracy of cyber language entities.Further,this paper uses the remote supervised relationship extraction model which combines syntactic dependency tree and ontology constraint layer to solve the problem that the result of relation extraction is not consistent with the fact.Based on the segmented convolutional neural network,the model introduces the dependency subtree of relation attributes to get the position weight of each word in the sentence.In the output layer,the domain ontology knowledge is introduced to constrains the extraction results to improve the accuracy of the extraction relationFinally,the application of knowledge map building network buzzword knowledge representation system,by analyzing the network buzzword reasoning to get the meaning of network buzzword,popular origin,nature,characteristics,such as for the detailed interpretation of the concept of network buzzword meaning,at the same time,through the knowledge representation system to each network buzzword with the phrase in detail the relationship between the visual display,can make people get the network buzzword deeper semantic understanding.To sum up,we boosts the algorithm of named entity recognition and relation extraction from the angle of knowledge graph explication,and the built up knowledge graph guarantee the accuracy of entity and relation attribute recognition and extraction of network terms on the premise of being consistent with the authenticity factuality.Ultimately,Using knowledge graph to design knowledge representation system for semantic analysis of network buzzwords The comparative experiment with other traditional model algorithms shows that the semantic analysis of network expressions based on the knowledge graph is more efficient.