Everybody uses the term "dataset". Is here any exact definition for it?

Asking because trying to draw an hierarchical diagram what show relations in open-data, comparing different legislations, especially EU and US.

So, in the wider view i'm interested mainly what is the relation between the next terms:

  • open-data & open-content & datasets
  • open-data & structured (csv,xml) & unstructured documents (memos, and such)
  • databases
  • documents
  • collections

and such terms...

Would be nice to get some longer explanation with some authoritative links.

1 Answer 1


No, there isn't.

The most in-depth analysis that I'm aware of is the 2011 paper by Renear, Sacchi and Wickett, "Definitions of dataset in the scientific and technical literature", in which they decided that there were four basic concepts that reoccurred (grouping, content, relatedness, purpose), but weren't completely consistent when dealing with different communities.

There have been other attempts to deal with the term 'data', which has significantly different meaning in the hard sciences, information science, and commputer science communities. See for instance, Ballsun-Stanton's 2010 paper, "Asking about data: Experimental philosophy of Information Technology".


As it's been so difficult to get fixed definitions, a few years ago, Todd King and I attempted to collect terms that we saw as problematic when attempting to talk about data systems with folks from other science disciplines. As it's been stable for more than a year now, we should probably try to publish it. (I wanted to provide cross walks to terms in OAIS and other vocabularies, so I'm the one holding it up). We specifically gave up on trying to define 'dataset' :

Please note that the terms dataset, data product and data series are not defined here. Although all refer to a grouping of data granules, the terms are used inconsistently across disciplines; in solar physics, a dataset is a collection of data products while in earth sciences, a data product is either a collection of similar datasets or a classification of datasets. These terms should be avoided, or clearly defined when used.

There's also an effort going on right now through the Research Data Alliance as part of their Data Foundation and Terminology Workgroup to come up with an overall data model ... I haven't been following their work as much as I should (and I still need to look at the draft they sent me a couple of weeks ago), but they should be releasing something for their plenary meeting at the end of the month.

  • once again one really great answer! ;) The doi links doesn't works, but the NASA (your) paper is excellent and your explanation too. Strange, nearly every second question here is about the "datasets" - everybody using a term what's meaning is defined only intuitively. Thank you. ;)
    – clt60
    Aug 7, 2014 at 19:46
  • @jm666 : the second DOI link should've worked ... I fixed the first one. I think for the most part we have a general concept of what 'dataset' means (some grouping of data), but not specific enough for someone to count how many they have in a consistent manner.
    – Joe
    Aug 7, 2014 at 19:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.