Emotional Attribution Enhanced Transformer for Movie Appreciation Classification

The accurate classification of film appreciation is essential for improving user experience, cultural understanding, and personalized recommendation systems. However, existing classification approaches face significant challenges due to the subjective nature of emotional expression, cultural diversity, and the multimodal composition of cinematic content. To address these challenges, this paper proposes an Emotional Attribution Enhanced Transformer (EAET) for movie appreciation classification. The model integrates a Graph-Transformer Network (GTN) to capture cultural semantic associations across diverse movie backgrounds, and a Cross-Modal Dynamic Gating (CMDG) mechanism to enhance the alignment of emotional information from textual and visual modalities. Experimental results on the Movie Trailers Dataset demonstrate that the proposed EAET model achieves a mean Average Precision (mAP) of 88.65% and an Accuracy (Acc@1) of 87.26%, outperforming existing state-of-the-art methods. Ablation studies confirm the complementary synergy between the GTN and CMDG modules. This work offers a novel computational framework for affective analysis in digital media, with potential implications for film classification, emotional computing, and cross-cultural content dissemination.