The Construction of Inheritance Network of Jiangsu Rural Intangible Cultural Heritage Skills Based on Knowledge Mapping and the Innovative Path of Locality

This study focuses on the construction of Jiangsu rural NRM inheritance network and proposes a knowledge graph-driven systematic solution. At the technical level, the NRM ontology framework (covering 99 types of entities and 199 attributes) is constructed based on the CIDOC CRM model, the RCBC named entity recognition model integrating four modules is designed, and the RoBERTa-Effg-Adv relational extraction model is proposed, which is trained by quintuple extraction with PGD confrontation. The RCBC named entity recognition model is realized on a self-constructed dataset with a F1 value of 83.11%, an improvement of 3.87 percentage points over the optimal baseline woBERT. The RoBERTa-Effg-Adv relational extraction model achieves an F1 value of 85.88%, which is significantly better than the BERT-BiLSTM-ATT model (+1.94%). At the application level, the analysis of 20 cultural heritage digital museums found that the quality of resource integration determines the effectiveness of communication (the PR value of a website with balanced inward/outward connectivity can reach 7, and the impact factor is 13.382). The study suggests that knowledge mapping can effectively structure the temporal, spatial, person and craft associations of ICH, while digital dissemination needs to be adapted to a multi-platform content strategy to achieve localized innovation.