A Study on Automated Issue Detection in Presidential Records through Topic Modeling
Korean Industrial Technology Convergence Society, Tae-Young Kim · Korea Industrial Technology Convergence Society · 2023
Presidential records, which capture the essence of social and political transformations, are closely related to understanding of important societal issues. Therefore, any endeavor to preserve and harness presidential records must also encompass concurrent consideration of related social issues. This study aims to propose a method that can periodically and automatically identify key issues, marking a departure from the one-off issue detection approaches previously employed. In view of this, we introduced an automated issue detection model specifically designed for presidential records, and verified its efficacy through an extensive collection of news media data from the past five years. The results were derived in several phases: acquisition of new data, preprocessing, optimal condition analysis for topic modeling, and topic modeling analysis. The newfound issues pertinent to presidential records were validated through visualization and case studies. Consequently, the study validated the usefulness of this approach by effectively identifying major issues concerning the presidential records.