The machine learning community currently has no standardized process for
documenting datasets, which can lead to severe consequences in high-stakes
domains. To address this gap, we propose datasheets for datasets. In the
electronics industry, every component, no matter how simple or complex, is
accompanied with a datasheet that describes its operating characteristics, test
results, recommended uses, and other information. By analogy, we propose that
every dataset be accompanied with a datasheet that documents its motivation,
composition, collection process, recommended uses, and so on. Datasheets for
datasets will facilitate better communication between dataset creators and
dataset consumers, and encourage the machine learning community to prioritize
transparency and accountability.
%0 Generic
%1 gebru2018datasheets
%A Gebru, Timnit
%A Morgenstern, Jamie
%A Vecchione, Briana
%A Vaughan, Jennifer Wortman
%A Wallach, Hanna
%A Daumé III, Hal
%A Crawford, Kate
%D 2018
%K annotation dataset ml
%T Datasheets for Datasets
%U http://arxiv.org/abs/1803.09010
%X The machine learning community currently has no standardized process for
documenting datasets, which can lead to severe consequences in high-stakes
domains. To address this gap, we propose datasheets for datasets. In the
electronics industry, every component, no matter how simple or complex, is
accompanied with a datasheet that describes its operating characteristics, test
results, recommended uses, and other information. By analogy, we propose that
every dataset be accompanied with a datasheet that documents its motivation,
composition, collection process, recommended uses, and so on. Datasheets for
datasets will facilitate better communication between dataset creators and
dataset consumers, and encourage the machine learning community to prioritize
transparency and accountability.
@misc{gebru2018datasheets,
abstract = {The machine learning community currently has no standardized process for
documenting datasets, which can lead to severe consequences in high-stakes
domains. To address this gap, we propose datasheets for datasets. In the
electronics industry, every component, no matter how simple or complex, is
accompanied with a datasheet that describes its operating characteristics, test
results, recommended uses, and other information. By analogy, we propose that
every dataset be accompanied with a datasheet that documents its motivation,
composition, collection process, recommended uses, and so on. Datasheets for
datasets will facilitate better communication between dataset creators and
dataset consumers, and encourage the machine learning community to prioritize
transparency and accountability.},
added-at = {2021-10-12T10:33:37.000+0200},
author = {Gebru, Timnit and Morgenstern, Jamie and Vecchione, Briana and Vaughan, Jennifer Wortman and Wallach, Hanna and Daumé III, Hal and Crawford, Kate},
biburl = {https://www.bibsonomy.org/bibtex/2ac74a8c28c1ec3e8ce7e99663018d6f8/asmelash},
description = {Datasheets for Datasets},
interhash = {26954fe6cef5f74dbbdf79dca2d97e28},
intrahash = {ac74a8c28c1ec3e8ce7e99663018d6f8},
keywords = {annotation dataset ml},
note = {cite arxiv:1803.09010Comment: Working Paper, comments are encouraged},
timestamp = {2021-10-12T10:33:37.000+0200},
title = {Datasheets for Datasets},
url = {http://arxiv.org/abs/1803.09010},
year = 2018
}