=================================================================================

In failure or risk analysis, e.g. in the oil and gas industry [1], historical data can be found in incident reports, which can contain many instances of past shortcomings or failures and can be used as learning experiences to prevent similar incidents from reoccurring. Companies use this knowledge to train their workers, and then the workers can study specific cases to identify the problems and to learn appropriate responses and countermeasures, and to improve the quality of the products.

Such reports contain different information, e.g. for the oil case, which is the location of the incident, time and date, name of the employer involved, contact information of the site contact, a general description of the incident, and even further details, root cause analysis, hazard and operability (HAZOP) studies, and basic risk ranking procedures such as risk matrices. To strengthen the existing system, the researchers had [1] applied a supervised machine learning [page4323] approach to accurately analyze and evaluate risk in incident reports. Such Artificial Intelligence (AI) [page4325] and Machine Learning (ML) [page4514] hold great promise for enhancing process safety management by visualizing data and recognizing patterns across big datasets in real-time, determining the most effective leading indicators, especially how they may relate to low-frequency high-consequence events, and prioritizing improvements to safety processes.

============================================

[1] Daniel Kurian, Fereshteh Sattari, Lianne Lefsrud, and Yongsheng Ma, Using machine learning and keyword analysis to analyze incidents and reduce risk in oil sands operations, Safety Science, 130(2020), 104873.