Toward Tracing Knowledge Flows in Martial Arts: Biographical Data and Interpersonal Contacts

This article presents an ongoing database construction effort for advancing evidence-based research on knowledge flows in Chinese martial arts. Martial arts, as embodied knowledge systems, intertwine the complexities of physical practice with ideological and sociocultural dimensions. Yet their histories remain elusive due to sparsity and divergence in documentation. To address these challenges, we propose developing a reliable knowledge database of martial arts practitioners, with a focus on biographical information and interpersonal contacts. In doing so, we experiment with a human-in-the-loop pipeline that combines prompt engineering with domain-specific semantics, iteratively evaluated and refined by domain experts. The pipeline extracts knowledge entities from curated historical corpora, both unstructured and semi-structured, and transforms them into structured datasets. By sharing the challenges, strategies, and preliminary outcomes, we introduce a pathway for organising knowledge within the underdocumented and heterogeneously documented martial arts historiography. This work lays a foundation for future analytics on the knowledge flows in martial arts, with potential applicability to other embodied traditions.