Using machine learning and keyword analysis to analyze incidents and reduce risk in oil sands operations

Journal article describing a supervised ML + keyword approach to improve incident reporting and analysis workflows. Using a large incident database, the authors build a library of labels/keywords and map them to standardized statements to automatically produce outputs such as a risk matrix placement, trend reports, prevention/mitigation strategies, and leading indicators for new incident entries. For PSM, this supports faster and more consistent identification of hazards and risk drivers from day-to-day incident data, helping organizations prioritize follow-up, strengthen barrier management, and feed learnings back into PHA and management review.