Construction and Application of a Knowledge Graph-Enhanced Generative AI System for Woodcarving Design in Intangible Cultural Heritage

Intangible cultural heritage (ICH) woodcarving lacks a structured knowledge system linking process knowledge, pattern connotations, and regional features. Generative artificial intelligence (AI) alone cannot ensure cultural semantic consistency, craft feasibility, or scene adaptability. This paper presents a woodcarving design system that integrates an ICH process knowledge graph. Multisource data are collected to build the graph. A retrieval-augmented mechanism with relationship constraints is introduced, forming a closed-loop process with five components: requirement analysis, knowledge matching, scheme generation, evaluation feedback, and weight correction. Experiments show that, compared with conventional generative AI, the knowledge graph–enhanced system achieves significantly higher levels of cultural semantic consistency, process feasibility, and design novelty, especially for complex tasks and specific scenes. Innovation is transforming ICH woodcarving knowledge into a core decision-making basis for generative AI and consequently offering a practical reference for digital preservation and design assistance.