Abstract: Aiming at problems of time-consuming, labor-consuming and limitation in constructing coal mine accident ontology by manual means, an automatic construction method of coal mine accident ontology was proposed. Firstly, taking text of coal mine accident case as data source, BP neural network is used to automatically extract ontology concept. Then, hierarchical relationship and non-hierarchical relationship between ontology concepts are automatically extracted by adopting hierarchical clustering method and association rules method. Finally, ontology concepts, relationship between concepts and examples of coal mine accident are visualized by use of Protégé ontology editor, so as to obtain coal mine accident ontology. The coal mine accident ontology by automatic construction is more comprehensive and can provide strong support for coal mine safety production work.