ChristineBasta's picture
Update README.md
b019079 verified
|
Raw
History Blame Contribute Delete
3.03 kB
metadata
configs:
  - config_name: en_es
    data_files:
      - split: train
        path: en_es/train.csv
      - split: validation
        path: en_es/val.csv
      - split: test
        path: en_es/test.csv
  - config_name: en_eu
    data_files:
      - split: train
        path: en_eu/train.csv
      - split: validation
        path: en_eu/val.csv
      - split: test
        path: en_eu/test.csv
  - config_name: es_eu
    data_files:
      - split: train
        path: es_eu/train.csv
      - split: validation
        path: es_eu/val.csv
      - split: test
        path: es_eu/test.csv
license: cc-by-sa-4.0
language:
  - eu
  - en
  - es

Dataset Description

This dataset contains a high-quality, real (non-synthetic) parallel corpus in the scientific and academic domain, covering Basque (eu), Spanish (es), and English (en).

Unlike synthetically generated datasets, this corpus is built from genuine, human-authored trilingual abstracts extracted from ADDI (Archivo Digital de Docencia e Investigación), the institutional digital repository of the University of the Basque Country (UPV/EHU). It is specifically designed to support the training, fine-tuning, and evaluation of Machine Translation systems and Large Language Models (LLMs) for specialized scientific text.

Source Data

The raw documents were scraped and curated from multiple academic collections within the ADDI repository. The texts are predominantly published under Creative Commons licenses.

Document Types Included:

  • Doctoral Theses
  • Master's Theses (Trabajos de Fin de Máster)
  • Bachelor's Theses (Trabajos de Fin de Grado)
  • Academic Journals and Journal Articles

Metadata Included: Each parallel record in the dataset is accompanied by rich metadata to facilitate targeted filtering. (Note: To ensure privacy, author and supervisor names have been explicitly anonymized).

  • Publication Date
  • Submission Date
  • URI (Original source link)
  • Language (The main language of the full publication)
  • License
  • Title
  • Department

The dataset consists of 3,294 perfectly aligned trilingual documents (yielding 3,294 tuples of Basque, Spanish, and English).

Metric Basque (EU) Spanish (ES) English (EN)
Total Documents 3,294 3,294 3,294

(Note: The lower average word count in Basque compared to Spanish and English, despite similar character counts, reflects the agglutinative morphological nature of the Basque language).

License

This data is released under Creative Commons CC-BY-SA license following the license of the scientific articles.

Contact

For questions or issues, please contact the ALIA / HiTZ team or open an issue on the Hugging Face dataset repository.

Citation

If you use this dataset in your work, please cite the ALIA project and the corresponding translation models used.

Funding

This work is funded by the Ministerio para la Transformación Digital y de la Función Pública - Funded by EU – NextGenerationEU within the framework of the project Desarrollo de Modelos ALIA.