Construction Of Medical Knowledge Graph Based On NLP And Application Of Intelligent Inquiry Platform

In recent years,the Internet has become an important channel for the public to obtain medical and health knowledge.The traditional online consultation platform based on passive search and matching not only cannot meet the individual needs of users,but also returns a large number of irrelevant search results,which results in users needing to spend a lot of time and energy to screen them one by one.Therefore,traditional online consultation platforms cannot meet the needs of users to accurately obtain medical knowledge.In this thesis,a knowledge graph in the medical field is constructed by integrating natural language processing technology to provide high-quality data support for the intelligent consultation platform.The work done in this thesis is as follows:(1)Based on dependency syntax,this thesis constructs an DP＿ABILSTM-CRF medical entity recognition model that optimizes the process of medical entity extraction.Based on the entity extraction baseline model BiLSTM-CRF,combined with dependency algorithm analysis,the grammatical structure and dependency relationship in the sentence are extracted to enrich the semantic features;and the attention mechanism is used to strengthen the current entity-related information and improve the accuracy of entity extraction.The experimental results show that the model in this thesis improves the performance of medical entity recognition and is beneficial to entity extraction in the medical and health field.(2)Based on the work of ABILSTM-CRF model for extracting medical entities,this thesis proposes a Mutil Att＿BiGRU model for extracting medical entity relations based on the dual attention mechanism.The BiGRU network is used to fully learn the semantic features,and the attention mechanism is integrated between the character layer and the sentence layer to optimize and improve the performance of the relation extraction model.The results show that the overall performance of the medical entity relationship extraction model proposed in this thesis has been greatly improved,and it is suitable for the relationship extraction process in the medical field.(3)Relying on the medical record data and CCKS series evaluation data provided by the Central Hospital of Xinyu Iron and Steel Group Co.,Ltd.,the proposed ABILSTM-CRF medical entity extraction model and Mutil Att＿BiGRU relation extraction method are applied to the medical knowledge extraction process to complete the extraction of medical entities and relationships.In terms of medical knowledge storage,Neo4 j graph database is selected to store the extracted medical knowledge after comparative analysis.Finally,a medical knowledge graph was successfully constructed based on the Neo4 j database,which can be queried and displayed through the Cyber language in the visual interface.(4)With the constructed medical knowledge graph as a high-quality knowledge source,an intelligent consultation platform is realized.Based on the medical knowledge extraction model and knowledge graph related technologies,it can accurately analyze user input,accurately perform knowledge retrieval,and immediately feedback the results on the frontend interface.The overall operation of the platform is good,it can accurately understand user input,and feedback results in real time,so as to meet the needs of users to obtain medical and health information.