Construction And Application Of Chinese Medical Knowledge Graph Based On CNKI

With the improvement of people's living standard,people pay more attention to biomedical health year by year,which is not only reflected in the growing demand for biomedical knowledge,but also in the continuous innovation of biomedical research in China.In recent years,the level of biomedical research in China has continued to improve,biomedical researchers produce a large number of biomedical literature every year.These data are large in quantity,complex in information and highly specialized.It is difficult for ordinary readers to understand their meanings.However,these documents contain a wealth of professional biomedical knowledge.The processing of these medical knowledge into structured information through text mining technology can bring great progress to the informationization of Chinese medical knowledge.The rapid development of natural language processing makes it possible to automatically extract biomedical entities and inter-entity relationships from biomedical literature.The extracted biomedical knowledge can be used to construct a biomedical knowledge graph to serve the intelligent development of biomedical data.Knowledge graph can transform unstructured data into structured data to promote understanding and application of knowledge units.In recent years,the construction and application of knowledge graph has received more and more attention from the industry,and a large number of enterprises try to apply them in business scenarios.This paper builds biomedical knowledge graph based on Chinese biomedical literature.The construction process is divided into four parts: biomedical knowledge acquisition,biomedical named entity recognition,entity relationship extraction and knowledge graph storage.This article focuses on Chinese biomedical related literature,so the literature data is mainly from China Knowledge Network(CNKI).In the stage of biomedical knowledge acquisition,this paper obtains biomedical literature summary data from CNKI by simulating the behavior of human click and performs data preprocessing.In the step of biomedical named entity recognition,BiLSTM+CRF is used as the model infrastructure,and the Attention mechanism is added to learn the dependence of each word on the full text.And the Chinese character radical feature is added to improve the recognition efficiency.In the stage of relationship extraction,this paper constructs a BiLSTM model that integrates the Multihop-Attention mechanism.First,use the BiLSTM model to learn the contextual features of each word;then,use the Multihop-Attention mechanism to learn multiple vector representations of sentences to provide more semantic information for the output layer.The rapid development of the Internet has made it an important channel for everyone to access information related to health care.However,when users use biomedical platforms or retrieval systems to query,they often need to browse a large amount of redundant or irrelevant information,it is always necessary to browse a large amount of redundant or irrelevant information.It is difficult to obtain the required information quickly and accurately.In order to solve the problem of low efficiency of traditional medical retrieval platform,this paper uses the constructed biomedical knowledge graph to assist the query process.The knowledge graph includes biomedical entities and relationships between entities.By visualizing,users can quickly locate the required information.Therefore,the biomedical knowledge graph constructed in this paper is applied in the biomedical literature retrieval system.The system adds functions such as medical entity markup,knowledge map visualization,and medical knowledge entity library based on the literature search function.The medical entity nouns of the literature abstract in the search results are marked by different colors,and the knowledge graph related to the query words are displayed in the form of network diagrams.The user can quickly obtain valuable information from the structured information,thereby improving the efficiency of retrieval.