Political Communication Construction in Short-Video Media Ecosystems: Participatory Dynamics and Mobilization Mechanisms

Abstract

Short-video platforms reorganize political communication by integrating compressed narratives, emotional cues, algorithmic visibility, and rapid audience response. This study develops a computational framework for examining how political messages on YouTube Shorts become participatory discussion and networked mobilization. The design combines video metadata, comment text, channel information, hashtag relations, and temporally aligned GDELT event records. BERTopic identifies narrative structures, VADER estimates affective orientation, and multilayer networks capture channel-topic, hashtag, and user-comment connections. A nonlinear mobilization index integrates action-oriented semantics, engagement, affective polarization, structural centrality, and event proximity. A reproducible benchmark run representing 2,847 validated videos and 248,613 comments produced six stable narrative themes with a mean topic coherence of 0.763 ± 0.031. Protest discourse generated the highest mobilization score (0.732 ± 0.156), while public-trust content generated the lowest score (0.518 ± 0.119). In the multilevel model, action-oriented semantics (β = 0.301 ± 0.028, p < .001) and network betweenness (β = 0.268 ± 0.044, p < .001) were the strongest predictors. The findings indicate that mobilization emerges from the coupling of narrative focus, emotional alignment, participatory density, and bridging positions within platform networks.