acukg

is a Knowledge Graph.

AcuKG is a comprehensive medical acupuncture knowledge graph that integrates acupuncture textbooks, official sources, clinical trials, biomedical literature, and ontology mappings to represent acupoints, anatomical locations, conditions, therapeutic indications, action targets, and evidence relationships.

License

SNOMED CT Affiliate License (inferred)

Homepage

acukg

Repository

GitHub

External IDs

No external IDs available

Product Summary

Contacts

Yiming Li

Github: yimingli99

Products

From this Resource
ID Name URL Category Format Description
acukg.csv AcuKG CSV tables AcuKG Graph csv Source CSV tables for AcuKG, includin...
acukg.rdf AcuKG RDF Turtle files AcuKG_RDF Graph ttl RDF Turtle representation of AcuKG re...
acukg.json AcuKG JSON files AcuKG_json Graph json JSON representation of AcuKG relation...
acukg.acupoint-vocabulary AcuKG local acupoint vocabulary local_acupoint_vocabulary.csv (14.5 KB) Product csv Multilingual acupoint vocabulary list...
acukg.mesh-conditions AcuKG MeSH condition list MeSH-Condition.csv (69.6 KB) Product csv List of MeSH descriptor IDs and terms...

Details

AcuKG structures medical acupuncture knowledge for retrieval, decision support, and ontology-aware analysis across acupoints, anatomy, conditions, evidence, and therapeutic indications.

Contents

AcuKG contains 1,839 entities (361 acupoints, 230 action targets, 447 indications, 404 anatomical locations, and 397 conditions) and 11,527 relations, mapped to 1,836 standard concepts in three ontologies (SNOMED CT, UBERON, and MeSH). Relation types include direction_of, part_of, distance_of, near_acupoint, located_near, has_clinicaltrials, has_pubmed, treats, and has_target.

Sources and construction

Acupoint codes, names, and locations were extracted from the WHO standard acupuncture point locations guideline; acupoint names, indications, and traditional actions were collected from the Acufinder.com web resource. Evidence links come from ClinicalTrials.gov and from PubMed clinical study articles retrieved with the query in pubmed_query.txt. Entities were obtained with rule-based and machine learning (CRF) named entity recognition and LLM-based relation extraction, with human review, then mapped to SNOMED CT, UBERON, and MeSH.

Access

The graph is distributed in the GitHub repository as CSV tables (AcuKG/), RDF Turtle (AcuKG_RDF/), and JSON (AcuKG_json/), along with relation subtype statistics. The authors also used a Neo4j instance of the graph for an LLM retrieval-augmented question-answering use case, but no public Neo4j endpoint is provided.

License

The repository does not declare a license. The license shown on this page is inferred from its most restrictive source (SNOMED CT).

Is this information incorrect or incomplete? Request an update.

Created: June 12, 2026 | Last modified: November 10, 2026