In recent years, discourse on appearance has become a social issue. This study attempted to categorize user attitudes using text mining and latent class models based on 3,185 free-form comments (narratives) on "lookism" posted on a free online question-and-answer service site (Yahoo! Answers). First, the free-form comments were processed using text mining to extract the presence or absence of 58 characteristic words such as "face," "plastic surgery," "discrimination," "romance," and "self-esteem." These were used as explicit variables, and a latent class model was applied, adopting a 5-class model based on the BIC criterion. As a result of the analysis, five types of attitudes were extracted: ① "Surface type" (32.5%), which has few specific mentions; ② "Cosmetic improvement type" (26.8%), which aims to improve appearance through plastic surgery and cosmetic procedures; ③ "Gender asymmetry type" (17.9%), which speaks of it as a disparity between men and women; ④ "Romance market type" (11.6%), which speaks of it in the context of romance and marriage hunting; and ⑤ "Physical self-dissatisfaction type" (11.1%), which speaks of physical complexes and low self-esteem. These results suggest that people's perceptions of lookism are not uniform but rather diverse, and will serve as foundational data for connecting individuals with mental health support.
