웹툰 섬네일의 시각적·심미적 특성과 관심도 간 관계 형태 탐색

In webtoon platforms where numerous contents are presented simultaneously, thumbnails function as critical visual cues that capture users’ attention and mediate content selection. However, prior studies have largely relied on fragmentary visual attributes or subjective evaluations, leaving the relational structure through which the visual and aesthetic characteristics of webtoon thumbnails are associated with user interest insufficiently examined. This study analyzes 996 webtoon thumbnails collected from major domestic platforms. Visual characteristics were quantified using the Quantitative Perceptual Index(QPI), while overall aesthetic quality was measured using MANIQA. The relationships with user interest were explored within a random forest–based nonlinear prediction framework. The results reveal that QPI, MANIQA, and the cumulative episode count exhibit distinct nonlinear patterns in relation to user interest. Meanwhile, the level of coupling among variables was not strongly observed within the scope of the present dataset and model specification. These findings suggest that analyzing the relational patterns of visual and aesthetic characteristics in disaggregated dimensions, rather than integrating them into a single composite measure, can provide useful insights for content analysis and platform strategy research.