An Empirical Study on AI-Driven Construction of Traditional Chinese Culture Knowledge Graphs

Abstract

To address the fragmentation, weak connectivity, and inefficiency in the digital preservation of traditional Chinese culture, this article proposes a systemic artificial intelligence framework for constructing and applying a cultural knowledge graph within a systems engineering paradigm. The framework integrates heterogeneous cultural data—including ancient texts, relic records, and historical events—through systematic data fusion. It employs a rule-enhanced deep learning model involving bidirectional encoder representations from transformers, a long short-term memory model, and conditional random fields to address semantic complexity, nested entities, and linguistic challenges in classical Chinese. A domain ontology capturing the people–event–object–emotion network in poetry enables structured organization and semantic consistency, while knowledge fusion ensures cross-source alignment. The graph, stored in Neo4j, is coupled with intelligent modules for question answering, spatiotemporal visualization, and sentiment analysis, enhanced by multimodal interaction.