数据集:

McGill-NLP/feedbackQA

预印本库:

arxiv:2204.03025

许可:

apache-2.0
中文

Dataset Card for FeedbackQA

? Read ? Code ? Webpage ? Demo ? Huggingface Dataset ? Discussions

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.

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

Apache 2.0

Contributions

McGill-NLP