Datasets:

Modalities:
Text
Formats:
json
Languages:
Basque
Size:
< 1K
ArXiv:
Libraries:
Datasets
pandas
License:
EusIE / README.md
OSainz's picture
Update README.md
2861b39 verified
---
license: apache-2.0
task_categories:
- token-classification
language:
- eu
pretty_name: EusIE
---
# Dataset Card for EusIE
<!-- Provide a quick summary of the dataset. -->
We present EusIE, the first Basque Event Extraction dataset. It inherits the annotation schema from [ACE 05](https://catalog.ldc.upenn.edu/LDC2006T06) (Automatic Contect Creation).
The dataset contains annotations for Named Entity Recognition, Event Extraction and Event Argument Extraction tasks as **segment level** (spans of 5 consecutive sentences).
## Dataset Details
### Dataset Description
<!-- Provide a longer summary of what this dataset is. -->
- **Curated by:** HiTZ Basque Center for Language Technology
- **Language(s) (NLP):** Basque
- **License:** Apache 2.0
- **Repository:** [link](https://github.com/MikelZubi/GrAL)
- **Paper:** [Zubillaga et al. (2024)](https://arxiv.org/abs/2404.06392)
## Uses
The dataset contains only a validation and test splits, and it is only intended to be used as evaluation.
## Citation
(LREC-Coling 2024 reference soon)
```bibtex
@misc{zubillaga2024event,
title={Event Extraction in Basque: Typologically motivated Cross-Lingual Transfer-Learning Analysis},
author={Mikel Zubillaga and Oscar Sainz and Ainara Estarrona and Oier Lopez de Lacalle and Eneko Agirre},
year={2024},
eprint={2404.06392},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
```