In recent years, functions such as comments and retweets on social media have made it possible to observe various public reactions and evaluations expressed online. This paper explores the online activities such as likes and replies on YouTube videos related to eleven Japanese Prime Ministers to investigate how former Prime Ministers are evaluated in online spaces. The study utilizes the YouTube Data API and Perspective API to measure the toxicity of comments and applies multiple regression analysis and wordcloud to explore factors affecting toxicity. The results show that for some Prime Ministers, higher toxicity in comments leads to more likes, while in other cases the opposite or different patterns appear.
