International Communication Effectiveness of Traditional Chinese Textile Culture Based on a Transformer Multimodal Fusion Model

With the rapid development of multimodal information transmission and intelligent semantic communication technologies, improving cross-modal consistency has become increasingly important for digital cultural dissemination and future communication systems. To address the problem of cross-modal semantic inconsistency in the international communication of traditional Chinese textile culture, this study proposes a culture-aware multimodal fusion framework based on Transformer architecture. Multilingual data collected from mainstream social media platforms are first utilized to construct a domain-specific knowledge graph covering craftsmanship, materials, patterns, and symbolic meanings. A cultural entity perception module is then introduced to enhance intra-modal feature representation, while a gated cross-modal attention mechanism dynamically aligns semantic information among text, images, and audio. Furthermore, a hierarchical fusion strategy generates unified communication representation vectors for communication effectiveness prediction. Experimental results demonstrate that the proposed framework achieves superior performance in cultural symbol recognition, cross-modal semantic alignment, and sentiment consistency, yielding an average Comprehensive Effectiveness Index (CEI) of 0.725 and significantly improving semantic fidelity and audience acceptance in international communication. By integrating knowledge-guided multimodal representation learning with adaptive semantic alignment, the proposed method provides an effective technical paradigm for intelligent cultural communication and offers valuable insights for multimodal semantic transmission, semantic communication architectures, and future wireless information dissemination systems.