Abstract: Compared with other production industries, metal mine is recognized as a high accident rate and the highest casualty rate due to the bad working environment. Therefore, safety production is the key concern of mining enterprises. With the attention of enterprises to safety problems and the increasing improvement of mine safety management system, many mines have established secure big data platform to effectively manage production and ensure the safety of underground operation, receiving the safety hazard information from daily safety inspection into the platform. However, due to the data of security risks are unstructured short texts with the operation of the enterprise, including the data recorded in the platform presents the characteristics of complex data content, large data scale, and non-standard data records. Moreover, due to the lack of an effective text analysis model, a small part of the security risk data is only used for simple analysis such as report analysis and data statistics, whereas more data is stored in a secure big data platform. Thus, the data did not play a guiding role in production, resulting in a waste of these valuable data resources. In order to explore the internal relationship between hidden danger data and the rule of hidden danger occurrence, based on big data analysis technology, this paper constructed a multi-dimensional analysis model of mine safety hidden danger. We analyzed the distribution law of hidden danger in two dimensions of time and space, used the topic mining model to classify hidden danger information, and obtained 13 hidden danger topics, using association rules to mine hidden danger. The model explores the internal relationship between different hidden dangers and uses an R programming language to visualize the above results. The results made full use of the mine hidden danger data and avoided the waste of data resources through the analysis and research of the hidden danger with a certain guiding value for preventing mine accidents.