Abstract Art-historical research has traditionally emphasized qualitative interpretation and visual analysis, but recent advances in machine learning have enabled the quantitative examination of visual and stylistic features that were previously difficult to analyze systematically. While large-scale databases have supported extensive analyses of Western artworks from various perspectives, comparable quantitative studies of Eastern visual traditions remain limited. In this study, we focus on Ukiyo-e, a traditional Japanese art form, as a representative case of Eastern art, and analyze artistic influence and structural change using a dataset of more than 11,000 digitized prints. The results show that compositional structure changed substantially over the nineteenth century: disruption increased from the 1820s to the 1880s and declined toward the end of the period. In contrast, stylistic properties remained stable, with little departure from null-model expectations. Taken together, these findings indicate that Ukiyo-e followed two distinct patterns of evolution, with noticeable shifts in compositional arrangements and persistent continuity in stylistic conventions.
Honna et al. (Tue,) studied this question.