Hugging Face's logo

Datasets:
ClementineBleuze
/
COCOA

Tasks:
Text Classification
Languages:
English
Size:
10M<n<100M
Tags:
nlp
claims
nlp4nlp
DOI:
License:
Dataset card Data Studio Files
xet
 Community

You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

By requesting access to this dataset, you agree to properly reference it in your work.

Log in or Sign Up to review the conditions and access this dataset content.

COCOA : A COrpus of Claims frOm NLP Articles

COCOA is corpus of sentences sourced from NLP papers and pre-prints published in English between 1952 and 2024. It contains additional annotations:

  • claim category labels reflecting the rhetorical function of sentences within papers (context, contribution, outline, result, impact, directions, limitation, non-claim); a part of these annotations was done manually, while the remainder labels were computed using a fine-tuned SciBERT model.
  • certainty annotations (sentence-level and aspect-level certainty) predicted using SciBERT models by Pei & Jurgens

Code related to the constitution, annotation and analysis of COCOA is available on GitHub.

This Huggingface dataset contains the data files of COCOA:

  • XML files of the papers included in COCOA
  • papers.csv: all the metadata about the papers included in COCOA
  • sentences-v1.csv: all the sentences of COCOA with annotations
  • sentences-v2.parquet: an updated version of sentences-v1.csv where some duplicates were removed
  • manual-sentences-v2.parquet: the subset of sentences from sentences-v2.parquet which were manually annotated

Parquet files should be loaded as follows: 

import pandas as pd
import fastparquet

sentences = pd.read_parquet("sentences-v2.parquet", engine = "fastparquet")

Contact

If you have requests or questions about this repository, please contact clementine.bleuze@univ-lorraine.fr.

Citation

You may reference this corpus using the following citation of our LREC paper: 

@inproceedings{bleuze:hal-05547842,
  TITLE = {{COCOA: Creation and Exploratory Investigation of a Corpus of Claims from NLP Articles}},
  AUTHOR = {Bleuze, Cl{\'e}mentine and Ducel, Fanny and Amblard, Maxime and Fort, Kar{\"e}n},
  URL = {https://inria.hal.science/hal-05547842},
  BOOKTITLE = {{LREC 2026 - International Conference on Language Resources and Evaluation}},
  ADDRESS = {Palma de Mallorca, Spain},
  ORGANIZATION = {{ELRA Language Resources Association}},
  YEAR = {2026},
  MONTH = May,
  KEYWORDS = {NLP4NLP ; ethics ; claims ; argumentative zoning},
  PDF = {https://inria.hal.science/hal-05547842v1/file/673_Paper-2.pdf},
  HAL_ID = {hal-05547842},
  HAL_VERSION = {v1},
}

COCOA is based on the ACL OCL corpus by Rohatgi et al.:

@Misc{acl_anthology_corpus,
  author = {Shaurya Rohatgi},
  title =  {ACL Anthology Corpus with Full Text},
  howpublished = {Github},
  year = {2022},
  url =  {https://github.com/shauryr/ACL-anthology-corpus}
  }

And it also used data from the ArXiv dataset on Kaggle:

@misc{arxiv_org_submitters_2024,
    title={arXiv Dataset},
    url={https://www.kaggle.com/dsv/7548853},
    DOI={10.34740/KAGGLE/DSV/7548853},
    publisher={Kaggle},
    author={arXiv.org submitters},
    year={2024}
}
Downloads last month
14
Total file size:
8.03 GB