Datasets:

ArXiv:
License:
feedbackQA / README.md
lizc's picture
Update README.md
157b664
|
raw
history blame
1.33 kB
metadata
license: Apache License, Version 2.0

Dataset Card for FeedbackQA

Dataset Description

Dataset Summary

FeedbackQA is a retrieval-based QA dataset that contains interactive feedback from users. It has two parts: the first part contains a conventional RQA dataset, whilst this repo contains the second part, which contains feedback(ratings and natural language explanations) for QA pairs.

Please find the full dataset here.

Languages

English

Dataset Creation

For each question-answer pair, we collected multiple feedback, each of which consists of a rating, selected from excellent, good, could be improved, bad, and a natural language explanation elaborating on the strengths and/or weaknesses of the answer.

Initial Data Collection and Normalization

We scraped Covid-19-related content from official websites.

Annotations

Who are the annotators?

Crowd-workers

Licensing Information

[More Information Needed]

Contributions

McGill-NLP