README.md

---
license: apache-2.0
task_categories:
- text-classification
language:
- en
tags:
- scify
- arxiv
pretty_name: WithdrarXiv
size_categories:
- 10K<n<100K
---
# Dataset Card for Dataset Name


## Who are the source data producers?

Authors on arXiv

## Who are the annotators?

ArXiv withdrawal comments are attributed to the authors of the respective papers. The categories come from prompting the Claude-sonnet model. 

#### Personal and Sensitive Information

We have taken measures to redact PIIs and filter rows to exclude any sensitive or personal information. Please see Section 8. of our paper.

## Bias, Risks, and Limitations

This dataset only contains retractions until Sep. 2024. For further information on risks, including ethical issues and the steps we take to mitigate them, please see Section 8 of our paper.



## Citation [optional]
**BibTeX:**

```
@misc{rao2024withdrarxivlargescaledatasetretraction,
      title={WithdrarXiv: A Large-Scale Dataset for Retraction Study}, 
      author={Delip Rao and Jonathan Young and Thomas Dietterich and Chris Callison-Burch},
      year={2024},
      eprint={2412.03775},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2412.03775}, 
}
```

## Dataset Card Contact

Delip Rao (@delip on huggingface)