Implementation of Latent Dirichlet Allocation (LDA) to Identify National Pride in YouTube Comments

Abstrak

Sports can trigger a strong sense of national pride, a psychological factor that contributes positively to collective behavior and societal cohesion. One key source of this pride is the success of athletes in prestigious international competitions. With the growing use of the internet, expressions of national pride are increasingly visible on social media, particularly in the comment sections of content related to national achievements. Previous studies on sports and national pride have mainly relied on survey methods. This study takes a different approach by analyzing expressions of national pride among Indonesian netizens in YouTube comments related to the success of Indonesian athletes in international sporting events. A case study was conducted using videos covering Indonesian achievements at the Paris Olympics 2024 in weightlifting, badminton, and rock climbing. Data were collected from 12 YouTube news videos using Mozdeh, yielding 11,972 comments, of which 10,610 were analyzed after preprocessing. The study applied Latent Dirichlet Allocation (LDA) to identify hidden topics, with the optimal number of topics determined as two based on coherence values. Term frequency–inverse document frequency (TF-IDF) was used for data representation. The results revealed two main themes: gratitude, reflecting thankfulness to God for achievements, and national pride, indicating public pride in Indonesia due to athletes’ success. These findings demonstrate that LDA effectively captures expressions of national pride in social media data and highlight the role of sports achievements in strengthening national pride online