Abstract: At present, coal mine gas explosion accident reports are in text record format. The specific cause information of the accident is contained in unstructured text, and there is a lack of professional accident dictionary. In order to solve the problem that manual analysis of accidents is time-consuming and laborious, low efficiency, and the analysis results are easily affected by subjectivity, and objectively extract the causes of accidents, we take “24Model of the cause of accident” as the theoretical basis of accident analysis, take the report of serious gas explosion accident in coal mine as the sample database, and construct the dictionary database of accident causes; with the help of information extraction technology, an accident cause information extraction algorithm based on rule pattern is proposed. The extraction results show that the extracted cause factors cover 90% of the manual analysis. This method provides theoretical support for rapid extraction of accident cause information in unstructured case base.