HuggingFace2RDF 0.1.1

HuggingFace Parser converting onnx repos to rdf

Hugging Face ONNX to RDF parser

A Python tool uses ONNX2RDF for converting HuggingFace Repositories with ONNX (Open Neural Network Exchange) files to RDF (Resource Description Framework).

---

✨ Features

• ✅ Parses ONNX model structure into RDF triples (nquads, turtle, trig, trix, jsonld, hdt)
• ✅ Automatic downloading of HuggingFace Files
• ✅ Keeps progress making it able to be stopped to later continue
• ✅ Download repositories randomfly, by most downloads or least downloads
• ✅ Use of threads and multiproccesing for paralization

---

⚙️ System Requirements

To use this tool successfully, the following components must be installed on your system:

• ✅ Python 3.13+
• ✅ Java 17+ (OpenJDK recommended)
  Used to run the internal RML Mapper JAR
• ✅ Node.js + npm
• ✅ @rmlio/yarrrml-parser
  Installed globally via npm install -g @rmlio/yarrrml-parser
• ✅ onnx-2-rdf (Pip)

You can also use the dockerfile (see below) which will prepare the enviroment.

---

📦 Installation Options

Option 1: Build from Source

Clone and install with PEP 621-compliant pyproject.toml:

git clone https://github.com/JorgeMIng/HuggingFace_ONNX2RDF.git
cd ONNX2RDF
pip install .

This installs the CLI command onnx-parser.

---

Option 2: Using Docker

git clone https://github.com/JorgeMIng/HuggingFace_ONNX2RDF.git
cd ONNX2RDF
docker build -t onnx2rdf .
docker run -it onnx2rdf

This Docker image includes:

• Python 3.13
• OpenJDK 17
• Node.js + npm
• @rmlio/yarrrml-parser
• ONNX2RDF + CLI (onnx-parser)
• HuggingFace2RDF + CLI (hugg-parser)

---

🚀 Usage

Command-Line Interface

hugg-parser num_repos [OPTIONS]

Positional Argument

Argument	Description
num_repos	Number of repos to parser from the list (all repos with ONNX tag) A value with (-1) means all repos.

Main Options

Option	Description
--target_path	Output directory for RDF files (default: rdfs). Can be absolute or relative.
--rdf_format	RDF serialization format: nquads (default), turtle, trig, trix, jsonld, hdt.

Pipeline Control

Option	Description
--work_folder	Changes the relative base folder for input/output (models, logs, RDF). Default to folder where the software is being called from
--num_threads	Number of threads to use (Default -1)

List Control

Progress list (progress_cache.json) stores repositories already done without error, repositories not able to be done (repo_id_error), with warnings (repo_id_warning), not done as it was stopped (repos_stopped)-> this repositories will have priority when executing again the program. and (repo_id_try_again) Has repositories which fail only because ONNX2RDF + repositories with warning. Repo_id_error are errors with Huggiface or unexpected errors

Option	Description
--order_method	Order method for selecting the models for the list (random (default), m_downs,l_downs)
--try_again	When try_again is executed only model on the list (repo_id_try_again) would be executed
--try_error	When try_error is executed only model on the list (repo_id_error) would be executed

---

Config Controls

A config file is present for changing values used for ONNX2RDF, there is a default config file, if a custom_parser.config is created the new values would be used to replace the final config set. The first time executing the program the file is created if it doenst exists in the workfolder.

• URIS is the most important category

Type	Option	Description
-----------	----------------------	-------------
URIS	resource_url	Namespace used for ONNX2RDF ahd HuggParser for modles :(http://base.huggingface.model.com/resource/)
URIS	hugginface_base	Namespace for Huggparser Ontology (yarrml file):(http://base.huggingface.model.com#)
-----------	----------------------	--------------
PARSER	debug	ONNX2RDF debug mode (True/False)
PARSER	cache	ONNX2RDF cache element o list of elements Valid Values: (all, load-model, pre-process, yamml2rml,mapping)
PARSER	work_folder	ONNX2RDF workfolder
PARSER	to_console	Activate ONNX2RDF console (True/False)
PARSER	models_folder	Folder for searching Huggiface files (Do not change)
PARSER	max_ram	RAM allocated for ONNX2RDF
PARSER	no_parsing	Desactivate Parsing (False) (Do not change)
-----------	----------------------	--------------
METRICS	file_name	Name of the metrics file to create
-----------	----------------------	--------------
CACHE	progress_file	Name of progress cache file
CACHE	cache_hugg_list	Huggiface model list cache file (file with all the ids, and metadata)
-----------	----------------------	--------------
METADATA	tmp_metadata_folder	Folder used to create metadata tmp files
METADATA	mapping_path	Folder to look up for the metadata yarrrml (looks up on instalation path) Not yet able to give custom yarrrml
METADATA	mapping_file	Name of metadata yarrrml (looks up on instalation path) Not yet able to give custom yarrrml
-----------	----------------------	--------------
LOGS	logs_folder	Folder for logs
LOGS	to_console	Activate HuggingFace2RDF console (True/False)

Example

hugg-parser 100 \
  --rdf_format turtle \
  --num_threads 4
  --order_method m_downs

This will:

• Parse the 100 most downloads repositories with turtle format
• Uses 4 threads

⚙️ Advanced Usage as a Library

Besides the command-line interface, ONNX2RDF also works as a Python library for programmatic integration.

Main HuggingFaceParser Class

The core class is HuggingFaceParser.

from HuggParser.HuggingFaceParser import HuggingFaceParser

parser = HuggingFaceParser()

# Set parameters
parser.set_rdf_format("nquads")
parser.set_number_threads(8)
parser.set_work_folder("results")

# Parse huggingface repositories
number_repos:int,try_again=False,try_error=False,order_method="random"
parser.run(
    number_repos=100,
    try_again=False,
    try_error=False,
    order_method="random",
)


## 📚 Related Resources


- [ONNX Format](https://onnx.ai/)  
  The Open Neural Network Exchange (ONNX) standard for representing ML models.

- [RDF Basics](https://www.w3.org/RDF/)  
  Introduction to the Resource Description Framework (RDF) by W3C.

- [SPARQL Tutorial](https://www.w3.org/TR/sparql11-query/)  
  Official SPARQL 1.1 Query Language specification and examples.

- [HuggingFace](https://huggingface.co)  
  Official HuggingFace Website.


## 🛠️ TODO

Improvements and future features can be found on [TODO.md (link)](TODO.md):

Feel free to contribute! Check out the [Issues](https://github.com/JorgeMIng/HuggingFace_ONNX2RDF/issues) tab for current tasks and discussions.

## 📄 License

This project is licensed under the terms of the Apache2.0 license.  
See [LICENSE](LICENSE) for more information.



## 🙌 Acknowledgments

- Built using the [ONNX](https://onnx.ai/) Python API.
- Built using the [HuggingFaceHub](https://github.com/huggingface/huggingface_hub) Official Hub API.
- RML Mapping powered by the [RMLMapper](https://github.com/RMLio/rmlmapper-java).
- YARRRML parsing supported via [@rmlio/yarrrml-parser](https://www.npmjs.com/package/@rmlio/yarrrml-parser).
- Developed by OEG (Ontology Engineering Group) of Polytechnic University of Madrid.



## 📑 Citation

If you use this software, please cite as:
 
```bibtex
@software{martin_izquierdo_2025_onnx2rdf,
  author       = {Jorge Martín Izquierdo},
  title        = {HuggingFace2RDF},
  version      = {0.1.1},
  date         = {2025-07-05},
  url          = {https://github.com/JorgeMIng/HuggingFace_ONNX2RDF},
  doi          = {10.5281/zenodo.15814658},
  license      = {Apache 2.0},
  affiliation  = {Universidad Politécnica de Madrid},
  keywords     = {ONNX, RDF, Semantic Web, Machine Learning},
  orcid        = {https://orcid.org/0009-0005-7696-8995}
}

📫 Contact

For questions, feedback, or contributions, feel free to reach out:

• GitHub Issues: https://github.com/JorgeMIng/HuggingFace_ONNX2RDF/issues
• Email: Jorge Martín Izquierdo – [e-mail]