Exploring the Interactivity of Cantonese Opera Videos on Youth-Centric Platforms

Using Python-based web scraping, we initially collected 4860 videos, narrowing them to 1916 through manual filtering by three coders. We extracted six quantitative variables (e.g., likes, shares) and seven categorical variables (e.g., author type, subtitle presence) for each video.

A binary interactivity variable was defined, categorizing videos into high and low interactivity based on the upper quartile of each quantitative metric. Using logistic regression, we analyzed how video characteristics influence the likelihood of high interactivity. Results indicate that factors such as author type, video tone, and subtitle presence significantly affect interactivity.