Student Presentations: INFORMS Annual Meeting 2019

SME-Miner: Utilizing Specialized Corpus Choices by Text Mining Application

Christie Nelson, Trefor Williams, John F. Betak, Kartheek Reddy, Jon McGuire, Ivan Mera, Alex Nguyen

A tool, SME Miner, has been created to analyze free-text fields in a somewhat novel manner: extracting subject matter expert (SME) identified keywords and utilizing the count of these keywords per-instance as an additional attribute in traditional machine learning methods. SME Miner automates the process of extracting SME-identified dictionary words from free-text, bins them by category, and creates a count-per-instance. We will show a proof-of-concept utilizing a railroad company’s maintenance log along with how one might do a second application utilizing a government agency’s background check data.