Model description

A question type classification model based on XLM-RoBERTa.

The question type classifier takes as input the question, and returns a label that distinguishes between boolean and short answer extractive questions.

The model was initialized with xlm-roberta-large and fine-tuned on the boolean questions from TyDiQA, as well as BoolQ-X.

Intended uses & limitations

You can use the raw model for question classification. Biases associated with the pre-existing language model, bert-base-multilingual-cased, may be present in our fine-tuned model, tydiqa-boolean-question-classifier.

Usage

You can use this model directly in the the PrimeQA framework for supporting boolean question in reading comprehension as in this example.

BibTeX entry and citation info

@article{Rosenthal2021DoAT,
  title={Do Answers to Boolean Questions Need Explanations? Yes},
  author={Sara Rosenthal and Mihaela A. Bornea and Avirup Sil and Radu Florian and Scott McCarley},
  journal={ArXiv},
  year={2021},
  volume={abs/2112.07772}
}

@misc{https://doi.org/10.48550/arxiv.2206.08441,
  author = {McCarley, Scott and 
            Bornea, Mihaela and 
            Rosenthal, Sara and 
            Ferritto, Anthony and 
            Sultan, Md Arafat and 
            Sil, Avirup and 
            Florian, Radu},
  title = {GAAMA 2.0: An Integrated System that Answers Boolean and Extractive Questions}, 
  journal   = {CoRR},
  publisher = {arXiv},  
  year = {2022},
  url = {https://arxiv.org/abs/2206.08441},
}