Processed official coal-mine safety event corpus and analysis outputs for risk-factor text mining

This versioned repository package supports the manuscript "Management implementation as a central risk signal in Chinese coal-mine accident investigation reports". It contains processed data tables, source manifests, evidence snippets, analysis outputs, figure source data, validation records, scripts, and package manifests for a text-mining study of official Chinese coal-mine safety materials. The primary analytical dataset (D1) comprises 86 formal coal-mine accident investigation reports. D2 comprises 172 additional records used for sensitivity and extension analyses. D3 comprises 304 official mining-safety event or case texts used only for robustness and external-validity checks, not as formal accident investigation reports. The package also contains locked manual validation summaries and audit records documenting author-confirmed validation decisions. Original official public-source web pages remain subject to the terms, availability, and copyright policies of the source agencies.