Tibetan–Qiang–Yi intangible cultural heritage dances convey rich cultural meanings through embodied movement, rhythm, costumes, and ritual settings. However, short-video dissemination often leads to decontextualization, fragmentation, and superficial learning, which weakens the inheritance of dance traditions. To address the difficulty of representing processual dance knowledge with a single medium, this paper proposes a multimodal knowledge graph framework for Tibetan–Qiang–Yi ICH dance. Using “movement–context unit” as the basic granularity, we construct temporal, spatial, and dynamic relational structures, and apply an evidence-chain mechanism to link multimodal materials with assertion nodes, thus preserving lineage variants and polyvocal interpretations. Pedagogical translation, which converts dance analytic language into executable training narratives, is integrated to support practical teaching. Furthermore, we design a qualitative evaluation pathway that operationalizes teachability, transmissibility, explainability, and correctability into observable indicators and assessment outputs. This framework provides a feasible approach for classroom training, community co-maintenance, and communication governance of ICH dance inheritance.

The data supporting the findings of this study are not publicly available because they contain community-sensitive cultural materials, restricted ritual knowledge, and field documentation subject to authorization and ethical constraints. Limited anonymized metadata or selected non-restricted materials may be available from the corresponding author upon reasonable request and with appropriate community permission.