Evolution and Transmission Trends of Film and Television Cultural Hotspots in the Short Video Ecosystem: A Time Series Analytical Approach

Abstract

Few studies have applied rigorous quantitative models to the temporal dynamics of hotspots. The cross-sectional design fails to capture the high-frequency, non-stationary features of the engagement data; The structural impact of the platform algorithm has not been formally integrated into the predictive model either. For this reason, utilize the 365 days of daily panel data for 2025 from Douyin to construct an all-encompassing time-series system in this study. Construct a composite Hotspot Index and apply the Augmented Dickey-Fuller test, ARIMA-GARCH model, Granger causality analysis and modified Bass diffusion model. The five stages of the lifecycle show significant conditional heteroskedasticity in the results. ARIMA models can perform well in forecasting, and Granger causality tests have shown that video volume, user engagement and key opinion leader participation all precede a rise in hotspot intensity. The modified Bass model has added a measure of algorithmic acceleration, and platform recommendation has reduced the traditional diffusion cycle by 62 per cent. The above results have been applied to content creators, platforms, and policymakers for culture.