A Study on the Influence Mechanism of User Interaction Behaviors on Movie Ratings

Abstract

This study investigates the relationship between user interaction metrics and movie ratings on Douban, focusing on the Top 250 films. Results reveal weak but statistically significant positive correlations between ratings and interaction metrics, with short reviews (r = 0.28) and the number of raters (r = 0.23) emerging as the strongest predictors. Regression analysis identifies short reviews (β=0.21) and raters (β=0.15) as primary drivers of high ratings, emphasizing the role of immediate audience engagement. Clustering analysis categorizes high-rated movies (ratings ≥ 9.0) into three interaction patterns: Extremely High Interaction, characterized by viral or timeless appeal; High Interaction, representing mainstream blockbusters; and Low Interaction, reflecting niche classics with limited modern traction. This study contributes to understanding the interplay between user behavior and film evaluation in digital communities, offering insights for algorithmic recommendation strategies.