siyue/SituatedQA

SituatedQA Dataset

This is a Context-Dependent Question Answering dataset, sourced from open-retrieval QA: Natural Questions, WebQuestions, TyDi-QA and MS-MARCO.

All of these datasets are in English and contain questions that are answerable by Wikipedia documents.

Question-context-answer triplets:

• Temp (train: 6009, dev: 3423, test: 2795)
• Geo (train: 3548, dev: 1398, test: 505).

Source

Please refer to github repo for source data.

Use

from datasets import load_dataset
dataset = load_dataset("siyue/SituatedQA","temp","train")

Example:

{
  "question": "where will the next summer and winter olympics be held",
  "id": "2098168902147822379",
  "edited_question": "where will the next summer and winter olympics be held as of 2021",
  "date": "2021",
  "date_type": "sampled_year",
  "answer": ["Japan and China"],
  "any_answer": ["Brazil and S. Korea", "Japan and China"]
}

Contact

For any issues or questions, kindly email us at: Siyue Zhang (siyue001@e.ntu.edu.sg).

Citation

@article{ zhang2021situatedqa,
  title={ {S}ituated{QA}: Incorporating Extra-Linguistic Contexts into {QA} },
  author={ Zhang, Michael J.Q. and Choi, Eunsol },
  journal={ Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP) },
  year={ 2021 }
}