Research On The Construction And Application Of Mental Health Knowledge Graph Of Medical Workers Based On Text Mining

Objective: With the advent of the era of big data,how to explore,organize and manage useful information in explosive data has become the focus of research.As one of the most important methods of knowledge management,knowledge atlas is widely concerned.Therefore,this study takes the literature in the field of mental health of medical staff as the basic data,and uses text mining and other methods to study the construction and application of the mental health knowledge graph of medical staff,so as to effectively sort out and present relevant information units for users,explore the known and unknown relationship between the mental health of medical staff and its influencing factors,and provide guidance for the protection and treatment of mental health of medical staff.Methods: According to the procedure of knowledge graph development,this paper has made improvements in the following aspects.(1)In the phase of domain concept and entity recognition,this paper proposes the Bio BERT-Bi LSTM-CRF model for named entity recognition for the first time.First of all,Bio BERT is used for text pre-training and word vector,which is more suitable for professional medical vocabulary in the biomedical field;In the feature extraction stage,Bi LSTM is used for feature extraction,which has a good fit for long and short term word vectors;Finally,the CRF layer can add some constraints to the final forecast tag to ensure its effectiveness.(2)In the relation extraction stage,this paper uses the method based on rule pattern matching and co-occurrence to extract the entity relationship.The pre-processed data is matched with the dependency syntax pattern that is applicable to the document data,and then combined with the results obtained by the co-occurrence method to obtain the entity relationship triple.(3)In the ontology construction stage,the standardization of mental health entities is realized through automatic matching of medical thesaurus and calculation of the similarity between entities and medical thesaurus.The seven-step method is used to construct the ontology of mental health influencing factors of medical staff,and the method based on rules and machine learning is used to classify the entities of influencing factors.Finally,the data is imported into Gephi to build a knowledge graph and realize visualization.(4)In the application phase of knowledge graph,calculate the knowledge graph index to carry out network analysis on the knowledge graph and find the key nodes in the knowledge graph.Through comparative analysis,the model is selected to complete the knowledge graph and explore the unknown relationship between entities in the knowledge graph.Results: The ontology of influencing factors of mental health of medical staff and the mental health knowledge graph of medical staff were constructed,and the entity query was realized.The network analysis of the network graph form of the knowledge graph shows that it has a small-world nature and the most influential risk factors include age,gender,work,marital status,education level,economic conditions,etc;The most influential protective factors include social support,psychological flexibility,psychological consistency,communication,psychological acceptance,etc.Finally,the Conv KB model is used to predict the link of knowledge graph,and the new relationship between risk factors,protection factors and mental health status is obtained.Conclusion: Through experiments,it is proved that the construction method of the mental health knowledge atlas of medical staff developed in this paper extracts key information from scientific literature in the field of mental health of medical staff,obtains the relationship between entities,and finally constructs the mental health knowledge atlas of medical staff.According to the network analysis of the knowledge atlas and the completion of the knowledge atlas,the key influencing factors are determined,and the potential relationship between the influencing factors and mental health problems is predicted,As the early foundation of digital medicine,it provides guidance theory for the early work of scientific researchers,and leads the way for future generations to develop the hidden resources of scientific literature.