pretty_name: HALvest-Geometric
license: cc-by-4.0
configs:
- config_name: en
data_files: en/*.gz
- config_name: fr
data_files: fr/*.gz
language:
- en
- fr
size_categories:
- 100K<n<1M
task_categories:
- text-generation
- fill-mask
task_ids:
- language-modeling
- masked-language-modeling
- graph-representation-learning
tags:
- academia
- research
- graph
annotations_creators:
- no-annotation
multilinguality:
- multilingual
source_datasets:
- HALvest
HALvest-Geometric
Citation Network of Open Scientific Papers Harvested from HAL
Dataset Description
- Repository: GitHub
Dataset Summary
overview:
This dataset is comprised of fulltext from open papers found on Hyper Articles en Ligne (HAL). Our dump is mostly english/french but gather papers written in 34 languages across 13 domains.
You can download the dataset using Hugging Face datasets:
from datasets import load_dataset
ds = load_dataset("Madjakul/HALvest-Geometric", "en")
Details
TODO
Languages
| ISO-639 | Language | # Documents | # mT5 Tokens |
|---|---|---|---|
| en | English | 442,892 | 7,606,895,258 |
| fr | French | 193,437 | 8,728,722,255 |
Graph
TODO
Considerations for Using the Data
The corpus is extracted from the HAL's open archive which distributes scientific publications following open access principles. The corpus is made up of both creative commons licensed and copyrighted documents (distribution authorized on HAL by the publisher). This must be considered prior to using this dataset for any purpose, other than training deep learning models, data mining etc. We do not own any of the text from which these data has been extracted.
Citation
TODO
Dataset Copyright
The licence terms for HALvest strictly follows the one from HAL. Please refer to the below license when using this dataset.