metadata

language:
  - en
pretty_name: CheckboxQA
tags:
  - Checkbox
license: cc-by-nc-4.0
task_categories:
  - document-question-answering

CheckboxQA

CheckboxQA consists of varied questions requiring interpretation of checkable content in the context of visually rich documents. Required answers range from simple yes/no to lists of values. The dataset is intended solely for evaluation purposes.

Example

Data format

Questions and answers for CheckboxQA follows DUE data format.

Evaluation

The evaluation metric for CheckboxQA is ANLS* (Average Normalized Levenshtein Similarity). You may evaluate your model's predictions using the evaluate.py script from Snowflake-Labs/CheckboxQA repository.

Citation

Please cite the following paper when using this dataset:

@misc{turski2025uncheckedoverlookedaddressingcheckbox,
      title={Unchecked and Overlooked: Addressing the Checkbox Blind Spot in Large Language Models with CheckboxQA}, 
      author={Michał Turski and Mateusz Chiliński and Łukasz Borchmann},
      year={2025},
      eprint={2504.10419},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2504.10419}, 
}

License

This dataset is intended for non-commercial research purposes and is provided under the CC BY-NC license.
Please refer to DocumentCloud's terms of service at https://www.documentcloud.org/home/ for information regarding the rights of the underlying document.