Ancient Chinese medical books contain valuable knowledge, but their contemporary technological relevance remains underexplored. This study aims to propose an empirical framework to assess their technological value through patent citation analysis supported by large language models (LLMs).
Guided by value theory and patent citation analysis, this study constructs a multidimensional evaluation system using three indicators: knowledge fusion, citation intent and citation position. LLMs are used for traditional Chinese medicine (TCM) entity recognition and citation intent classification, while semantic similarity between patent titles and cited texts is calculated using the GanymedeNil/text2vec-large-Chinese model. The CRITIC method determines indicator weights.
Empirical findings reveal that knowledge fusion and citation intent are the most influential indicators, with citation position playing a supplementary role. Patent citations referencing TCM knowledge demonstrate a hierarchical structure, with 72.9 % exhibiting low levels of knowledge integration. The technological value of ancient TCM books is primarily manifested through providing foundational support and inspiring innovation in modern TCM applications, paralleling patterns observed in contemporary scientific literature. Furthermore, notable disparities exist among different texts: Systematic Differentiation of Warm Diseases ranks highest in technological impact, and books focusing on warm diseases and gynecology outperform those related to health preservation.
This study integrates LLMs with patent citation analysis to overcome limitations of traditional approaches, offering a scalable method for evaluating the technological value of ancient Chinese medical books in contemporary innovation contexts.
