Term: Data Element Class:  
 vernacular   (0%)
Created 20 September 2013
Last modified 20 September 2013
Contributed by Gary Berg-Cross


Definition: The most elementary unit of data that can be identified and described in a dictionary or repository. A data element cannot be subdivided.

What does "cannot be subdivided" mean here? Consider for example a video sequence. May this be considered a data element (e.g., in a collection of videos) despite the fact that video files can typically be subdivided into image frames and audio tracks? Please clarify. Submitted 14 April 2016
by Raphael
Video's particularly bad as you can both subset as you mentioned by time (individual frames, or a range of frames), by channel (audio vs. video vs. subtitles), or even spatially (the whole time series, but only the upper left quadrant, or only tracking around a given feature moving with the frames). We *might* be able to qualify this definition to state that the data system within which the item is registered can't do the subdivision, but I'd have to see enough context via the other definitions to see how problematic this would be.

See a talk I gave a couple of years back for some of the issues of aggregation & subsetting : http://vso1.nascom.nasa.gov/vso/misc/FRBR-aggregates-code4lib.ppt
Submitted 14 April 2016
by Joe
I suggest that the definition be changed to "The most elementary unity of data that is identified and described in a repository or catalog. Data elements may be further subdivided through the use of ancillary tools. Also known as a "granule" in NASA's and NOAA's vernacular for data from satellites. "

With this sort of definition we could then use a few examples such as the video example above....
Submitted 14 April 2016
by Ruth
I agree with Ruth here. I think the situation is not uncommon (i.e., composite file/data formats). In neuroscience DICOM, NIFTI, GIFTI, CIFTI, NEO and the likes come to my mind but I've chosen video to start the discussion as it is more generic. Submitted 14 April 2016
by Raphael
I also like Ruth's proposal. Like many of the concepts we're dealing with (metadata vs. data, data object vs. data citation, data element, data, dataset), in the end the definitions have an intention component. In this case, a data element is the smallest unit of data that some community chooses to consider as an entity (http://seaice.herokuapp.com/term=1041) for some purpose. We still need to decide if there different concepts labeled by 'data element', 'datum', 'data item'. Submitted 14 April 2016
by Stephen
Perhaps 'data element' should denote the 'elementary unit of data' concept, and 'data item'should be a more general concept of 'an identified unit of data'. Both concepts are relative to some context/purpose/community. Submitted 14 April 2016
by Stephen
I like the distinction between 'data element' as an atomic ' unit of data' concept, and of 'data item' as a more general concept of some identified unit of data meaning that it is unitary for some purpose as in a video use or was the original unit in which it was captured but might be broken into pieces later.. Submitted 14 April 2016
by Gary