TCM-MKG Reorganized Graph Export

本仓库是对原始 TCM-MKG V1.0 的格式整理与图结构导出，不是原始数据集的官方发布，也不声称拥有底层知识内容。

Dataset Description

This repository contains a Hugging Face-compatible, reorganized graph export derived from the publicly available Traditional Chinese Medicine Multidimensional Knowledge Graph (TCM-MKG V1.0).

JX-Lab performed format conversion and schema normalization to make the resource easier to load for knowledge graph and machine learning workflows. The repository retains only the reorganized nodes.tsv and edges.tsv outputs. Copies of the original Zenodo tables and documentation are not distributed in this public repository.

The legacy directory name original_kg/ is retained for API and link compatibility; its files are reorganized exports generated by JX-Lab, not the untouched original source tables.

Files

• original_kg/nodes.tsv: 369,912 normalized node records
• original_kg/edges.tsv: 48,849,793 normalized relationship records

The exported schema preserves provenance fields such as source_table, evidence, and attributes_json so records can be traced back to source-table identifiers.

Original Source And Attribution

The underlying knowledge content comes from:

Traditional Chinese Medicine Multidimensional Knowledge Graph (TCM-MKG V1.0)

• Original authors: Zeng, Jingqi; Jia, Xiaobin
• Publisher: China Pharmaceutical University
• Original source: https://doi.org/10.5281/zenodo.13763953
• Original release year: 2024

TCM-MKG integrates information from multiple biomedical, terminology, pharmacopoeia, and traditional Chinese medicine databases. Rights in that source content remain with the original authors and the respective upstream rights holders.

What JX-Lab Changed

JX-Lab's contribution is limited to technical organization and format conversion, including:

• transforming source-specific tables into unified node and edge tables
• normalizing entity identifiers and relationship fields
• retaining source-table provenance and structured attributes
• preparing Hugging Face dataset configurations for streaming and loading

No claim is made that JX-Lab collected the original biomedical evidence or created the original TCM-MKG knowledge resource.

License And Third-Party Rights

This repository is marked license: other because the original Zenodo record does not specify a formal license in its Rights/License field. JX-Lab does not apply a new blanket CC BY, MIT, or other license to the underlying source data.

Use of this reorganized export remains subject to:

1. the terms stated by the original TCM-MKG authors;
2. the licenses and terms of each incorporated upstream database or terminology;
3. applicable copyright, database-right, contract, attribution, and research-use requirements.

Public availability does not by itself remove those obligations. Users must verify that their intended use, especially redistribution or commercial use, is permitted.

Usage

from datasets import load_dataset

nodes = load_dataset("JX-Lab/TCM-MKG", "nodes", split="train")
edges = load_dataset("JX-Lab/TCM-MKG", "edges", split="train", streaming=True)

Streaming is recommended for the edge table.

Citation

Please cite the original dataset:

@dataset{zeng_jingqi_2024_13763953,
  author       = {Zeng, Jingqi and Jia, Xiaobin},
  title        = {Traditional Chinese Medicine Multidimensional Knowledge Graph (TCM-MKG)},
  year         = {2024},
  publisher    = {China Pharmaceutical University},
  version      = {V1.0},
  doi          = {10.5281/zenodo.13763953},
  url          = {https://doi.org/10.5281/zenodo.13763953}
}

When describing this repository, please identify it as a reorganized Hugging Face export and indicate that JX-Lab performed the format conversion.

Rights-Holder Notice And Takedown

本仓库尊重原作者及各上游数据权利人的权益。如权利人认为本整理版包含未经适当授权、不应公开或署名不准确的内容，请通过本仓库的 Hugging Face Community 页面联系 JX-Lab，并提供相关权利和文件范围说明。收到并核实后，将及时删除、限制访问或修正相关内容。

This notice is intended to provide a practical correction and takedown channel; it does not replace any permission required for hosting or use.

Disclaimer

This dataset is provided for research and technical evaluation. It does not constitute clinical diagnosis, treatment, prescription, or medication advice. JX-Lab makes no warranty regarding completeness, accuracy, title, or non-infringement.