ADVANCED APPLICATION OF TEXT ANALYTICS IN MSHA METAL AND NONMETAL FATALITY REPORTS

摘要

Advanced data science techniques have been applied in a verity of fields to improve health and safety and reduce accident and fatality rates. Similar to other industries, the mining industry requires the adoption and implementation of modern research methods to continue boosting safety in mine operations. Text analytics, in general, is one of the essential methods used to analyze unstructured data that is in text format. Using text mining and Natural Language Processing (NLP) techniques to extract patterns with incidents and identify insightful information from contents and reports is extremely valuable to produce actionable recommendations and strategies for applying text mining and NLP to accident text data. In this research investigation, advanced text mining techniques and NLP were applied to the U.S. Mine Safety and Health Administration (MSHA) final fatality reports in metal and nonmetal operations (surface and underground) to perform preprocessing, Exploratory Data Analysis (EDA), and various statistics to discover insightful word relation patterns in the reports. For this purpose, the final fatality reports from 2010 through 2017 were collected and cleaned. In addition, topic modeling was done to group the reports with similar underlying themes or topics.