This paper surveys 60 English Machine Reading Comprehension datasets, with a view to providing a convenient resource for other researchers interested in this problem. We categorize the datasets according to their question and answer form and compare them across various dimensions including size, vocabulary, data source, method of creation, human performance level, and first question word. Our analysis reveals that Wikipedia is by far the most common data source and that there is a relative lack of why, when, and where questions across datasets.
@article{arxiv.2101.10421,
title = {English Machine Reading Comprehension Datasets: A Survey},
author = {Daria Dzendzik and Carl Vogel and Jennifer Foster},
journal= {arXiv preprint arXiv:2101.10421},
year = {2021}
}
Comments
Will appear at EMNLP 2021. Dataset survey paper: 9 pages, 5 figures, 2 tables + attachment