Research And Application Of Knowledge Mapping Technology For Specific Domain

Knowledge graph are divided into open domain knowledge graph and domain-specific knowledge graph.Domain-specific knowledge graph have broad research and application prospects.This article is devoted to the research and application of knowledge graph construction techniques in specific areas,and uses construction library resources and test resources as corpus.Extract knowledge of construction concepts and build knowledge graph and apply them.The main research contents of this article include:1.Research how to extract professional domain concepts from linguistic texts,propose a concept extraction algorithm based on conditional random field and information entropy,and how to extract features using the condition random field to train,and then use the information entropy to improve the extraction concept.This method effectively improves the accuracy and recall of concept extraction.A concept extraction algorithm based on BILSTM + CRF and information entropy is proposed to extract features automatically.The output layer use condition random field to label sequences and use information entropy to edit.This method can effectively improve the recall rate of concept extraction,reduce the extraction cost,and increase the automation degree of extraction.2.Research how to identify the relationship between the concepts of the construction field,face the concept data to extract concept pairs,use the upper and lower position and synonym relationship recognition algorithm based on the conditional random field,through the artificial extraction of upper and lower position and synonymous word features to train identification model.By defining the evaluation index,the comparison with the baseline method demonstrate the superiority of the method.3.Use the concepts and relationships of the construction to complete the construction of the knowledge graph of the construction field.The recommendation system based on the knowledge graph of the construction field is designed and implemented.From the user's wrong question,the concepts and questions are recommended to users based on collaborative filtering and content to achieve personal learning.