Datasheet for Datasets: Definition

Ask about Datasheet for datasets

Answers are drawn from this dictionary entry and the rest of the CASRAI corpus, with a link to every source.

Answers are AI-generated from CASRAI’s own published pages and can be wrong, so check the linked sources before relying on one; your question is logged without personal data — never sold, never used to train a third-party model — to show us what CASRAI is missing, so please do not type personal or confidential details. How we use this

Gebru et al. (2021) proposed datasheets to make dataset provenance and limitations visible to downstream model builders. Topics covered include consent and licensing of subjects, sampling and labelling procedures, demographic composition, known biases, and recommended/cautioned uses. Datasheets are complementary to model cards.

References

Gebru et al., 'Datasheets for datasets' (Communications of the ACM, 2021).

Frequently Asked Questions

Who introduced datasheets for datasets, and where was the concept published?

The concept was proposed by Gebru et al. in a 2021 paper, ‘Datasheets for datasets,’ published in Communications of the ACM, to make dataset provenance and limitations visible to downstream model builders.

What does a datasheet for datasets document?

It is a structured document accompanying a machine-learning dataset that records its motivation, composition, collection process, pre-processing, intended uses, distribution, and maintenance, modelled on datasheets used for electronic components. In practice this covers consent and licensing of subjects, sampling and labelling procedures, demographic composition, known biases, and recommended or cautioned uses.

How is a datasheet for datasets different from a model card?

A datasheet documents the dataset itself, while a model card documents the model trained on it. The two are complementary, not interchangeable.

Does a dataset README count as a datasheet for datasets?

No. A README that only lists file formats and column descriptions does not qualify. A datasheet for datasets additionally covers consent, sampling and labelling procedures, demographic composition, known biases, and recommended or cautioned uses.

What’s an example of a dataset with a datasheet for datasets?

Examples include a face-recognition benchmark distributed with a datasheet listing its demographic composition, consent procedures, and recommended uses, and an NLP corpus accompanied by a datasheet documenting its source domains and crawling rules.

Use in your systems

JATS XML <role> element

xml

<role vocab="credit"
      vocab-identifier="https://casrai.org/dictionary/"
      vocab-term="Datasheet for datasets"
      vocab-term-identifier="https://casrai.org/dictionary/term/datasheet-for-datasets" />

Schema.org DefinedTerm (JSON-LD)

json

{
  "@context": "https://schema.org",
  "@type": "DefinedTerm",
  "@id": "https://casrai.org/dictionary/term/datasheet-for-datasets",
  "name": "Datasheet for datasets",
  "identifier": "https://casrai.org/dictionary/term/datasheet-for-datasets",
  "description": "A structured document accompanying a machine-learning dataset that records its motivation, composition, collection process, pre-processing, intended uses, distribution, and maintenance, modelled on electronic-component datasheets.",
  "inDefinedTermSet": "https://casrai.org/dictionary/domain/ai-ml-research-outputs#set",
  "url": "https://casrai.org/dictionary/term/datasheet-for-datasets",
  "sameAs": [
    "dataset datasheet"
  ],
  "license": "https://creativecommons.org/licenses/by/4.0/",
  "publisher": {
    "@id": "https://casrai.org/#organization"
  },
  "dateModified": "2026-08-15T04:41:15",
  "inLanguage": "en"
}