Communication alienation and value reconstruction of literary and artistic criticism in the era of short videos: Taking film and television review exts as an example

Existing research is largely limited to qualitative exploration of the impact of short videos on literary and artistic criticism, lacking quantitative empirical evidence and causal evidence for the core mechanism of "communication alienation".The purpose of this study is to provide a computational framework that can successfully dissect the mechanism of alienation through dynamic topic modeling and multimodal sentiment divergence method.This framework first measures the sentiment of review text using a BERT (Bidirectional Encoder Representations from Transformers)-based model and leverages pre-trained automatic audio and video analysis models, VGG16 and OpenSmile, to extract high-dimensional sentiment information from audio and video data.Next, a cross-modal attention mechanism is then used to calculate the relative weight of audiovisual sentiment to textual sentiment, quantifying the degree of sentiment divergence in the "form drives content" model.Furthermore, a dynamic topic model (DTM) is used to track the temporal decay of the topic depth in comment texts.Granger causality tests are then conducted with video communication metrics to empirically demonstrate the causal relationship between alienation features and traffic incentives.Experimental results show that the mean completion rate increases from 0.28 in the low-alienation range to 0.49 in the high-alienation range, confirming the driving effect of formal expression on communication effectiveness.Multimodal features indicate that the visual modality has a weight of 0.45 in predicting completion rate, significantly higher than the 0.25 weight of the text modality, reflecting the alienated nature of "form overwhelms content".This study provides quantifiable analysis and a solid theoretical basis for understanding the changing form of literary and art criticism in the algorithmic era, and also provides an operational governance path for platform algorithm optimization and value reconstruction of the comment ecosystem.