scilons/SciLaD-all-xml-v1

Dataset Card for SciLaD

SciLaD is a novel, large-scale dataset of scientific language constructed entirely using open-source frameworks and publicly available data sources. It comprises a curated English split containing over 10 million scientific publications and a multilingual, unfiltered TEI XML split including more than 35 million publications. We also publish the extensible pipeline for generating SciLaD.

Dataset Details

In this repository we share the full, multilanguage TEI-XML version of the dataset. It was processed with Grobid 0.8.1. We plan to release updates at each new Grobid release.

The dataset is structured as follows:

• there are ~3000 partitions of the dataset
• each partition contains three files:
  • tei.zip containing the TEI-XML version of each article, one per file
  • metadata.zip contains the metadata information of each article
  • map_jsonl.gz contains the mapping between metadata and tei

Dataset Sources

• Repository: https://github.com/scilons
• Paper: https://arxiv.org/abs/2512.11192

Citation

The paper has been accepted at LREC 2026, the citation will be updated with the Camera ready version as soon as possible.

BibTeX:

@misc{foppiano2025sciladlargescaletransparentreproducible,
      title={SciLaD: A Large-Scale, Transparent, Reproducible Dataset for Natural Scientific Language Processing}, 
      author={Luca Foppiano and Sotaro Takeshita and Pedro Ortiz Suarez and Ekaterina Borisova and Raia Abu Ahmad and Malte Ostendorff and Fabio Barth and Julian Moreno-Schneider and Georg Rehm},
      year={2025},
      eprint={2512.11192},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2512.11192}, 
}