Generative AI-driven Semantic Compression and Trustworthy Transmission Mechanisms for Cross-cultural Digital Heritage

In the context of intensified digitalization and globalization, cross-cultural transmission of cultural heritage faces dual challenges of semantic compression and trust construction. Existing methods often fail to balance semantic fidelity, transmission efficiency, and cultural adaptability, limiting the effective flow of digital heritage across diverse contexts. This study proposes a Generative AI-based semantic compression and trustworthy transmission framework, integrating the T5 language model, semantic alignment algorithms, and a multidimensional trust model to optimize compression and delivery of digital heritage in multilingual settings. Tests were conducted on open-source UNESCO, Europeana, and Google Arts & Culture data in English, Chinese, Arabic, and French environments. Experimental results show the new approach significantly outperforms traditional approaches in semantic fidelity, compression ratio, cultural adaptability, and explainability, achieving an average 34.8% improvement in cross-cultural transmission trustiness. This work accomplishes more than verifying the technical potential of Generative AI in semantic compression but offers an expandable smart solution to global dissemination of digital cultural heritage.