Datasets:

GSMA
/

telecom-kg-rel19

license: cc-by-4.0
task_categories:
  - text-retrieval
  - text-ranking
  - text-generation
  - text-classification
  - question-answering
language:
  - en
tags:
  - telecom
  - 3gpp
  - knowledge_graph
  - 6g
  - standards
  - llm
  - rag
  - information_extraction
pretty_name: 3GPP Rel-19 Telecom Knowledge Graph
size_categories:
  - 100K<n<1M

3GPP Rel-19 Telecom Knowledge Graph (KG + Chunks)

Dataset Summary

This dataset, developed by Khalifa University Research Institute for Digital Future (KU-DF), contains a large-scale telecom-domain knowledge graph built from 3GPP Release 19 specifications, together with the underlying text chunks used for extraction and grounding.

It supports research and development in:

Telecom and networking
Knowledge graphs
LLM-based reasoning over standards
Retrieval-Augmented Generation (RAG)

Dataset Structure

The dataset includes three main components:

Knowledge Graph (graphml):
- Nodes: telecom entities (concepts, parameters, identifiers, protocols, references)
- Edges: semantic relations with provenance
Text Chunks (jsonl):
- Fine-grained chunks extracted from 3GPP specifications
- Used as grounding for KG nodes and edges
Entity Mapping (json):
- Maps entities to chunk IDs and source files

Data Statistics

Nodes: 21,540
Edges: 31,718
Unique entities: 6,044
Text chunks: 896,453

Data Cleaning

Cleaning was limited to structural normalization:

Removal of NULL-only or empty identifiers
Removal of isolated nodes with no semantic content
Exact duplicate node and edge removal
No semantic pruning of telecom concepts

Intended Uses

Telecom standards analysis
Knowledge graph analytics
LLM-assisted exploration of 3GPP specifications
RAG systems for telecom documentation
Research on information extraction from standards

Limitations

The dataset is limited to 3GPP Release 19
Some entities may lack textual descriptions and are intended for later enrichment

License

This dataset is released under the Creative Commons Attribution–NonCommercial 4.0 International (CC BY-NC 4.0) license.

The dataset is derived from publicly available 3GPP specifications.
Users must provide appropriate attribution and may not use the dataset for commercial purposes.

Additional Information

For full details on construction and usage, see the accompanying README.md.

Acknowledgement

Compute resources for building and experimenting with the knowledge graph were provided by the ITU AI for Good AWS Sandbox.

Citation

If you use this dataset, please cite the dataset repository

@dataset{kudf_rel19_telecom_kg_2026,
  title     = {3GPP Release 19 Telecom Knowledge Graph},
  author    = {Yang, Yuzhi and Bariah, Lina and Lu, Yuhuan and Debbah, Merouane},
  year      = {2026},
  publisher = {Hugging Face},
  organization = {Khalifa University Research Institute for Digital Future (KU-DF)},
  abstract  = {A large-scale telecom knowledge graph constructed from 3GPP Release 19 specifications, together with text chunks for grounding, provenance, and retrieval-augmented generation (RAG).},
  url       = {https://huggingface.co/datasets/otellm/telecom-kg-rel19},
  note      = {Part of the GSMA Open Telco Assets Initiative.}
}