Abstract: 【Purpose/significance】Knowledge representation and reasoning for treatment of public health emergencies through a knowl?
edge graph to drive assisted treatment applications and promote the effective integration and utilization of medical knowledge【. Method/process】Taking the COVID-19 epidemic as an example, firstly, BiLSTM-CRF and TF-IDF are used to identify the symptom vocabu? lary in the case information, expand the symptom words by HIT synonym forest, and extract reusable concepts and hierarchical rela? tionships from online ontologies using lexical similarity. Then medical knowledge graphs are constructed based on RDFlib ontology au? tomation strategy. Based on the latest version of the treatment regimen, SWRL rules are designed to perform case typing, medical plan generation and visualization of treatment knowledge by inference rules and neo4j.【Result/conclusion】Based on 68 SWRL inference rules, 6 types of persons and a combination of Chinese and Western medical treatment measures based on 13 main types of treatment plans were inferred, the integration of data and knowledge of symptoms and treatment regimen are also achieved.【Innovation/limita? tion】The actual case symptoms and online ontology knowledge are fused, and the treatment regimen is used as the basis for the infer? ence rules of the knowledge graph to realize medical knowledge representation and inference from a digital intelligence perspective. Further expansion of case characteristics and actual treatment measures is still needed to provide a richer basis for intelligent assisted diagnosis and treatment.