Abstract: In order to scientifically prevent and control the accident of coal mine gas, it is necessary to clarify the causes of coal mine gas accidents so as to take corresponding control measures. Based on the text mining method and R language, 126 domestic investigation reports of coal mine gas accidents from 2012 to 2020 were collected for word processing, feature dimension reduction and word cloud visualization, then the network structure diagram was constructed, and the social network centrality analysis and core edge structure analysis were carried out. The results show that there are 4 key causes and 27 general causes among the 31 causes obtained from text mining. According to the network density of each cause, the core area and edge area in the network are divided.