What is metadata?

NOTE HXA7241 2011-08-01T10:44Z

It is usually described as ‘data about data’. But there is a slightly different angle to see it from – not so much ‘above’ but ‘beside’.

Metadata is often thought of as ‘data that describes other data’: which leads to the thought of meta-metadata, data describing metadata, and so on, as if up a series of levels. But this is misleading: there is no hierarchy, there is no sense of each standing a level above the last. Metadata is better thought of as flat – as an association that adds structure and is separable usage-wise.

Metadata adds more structure to its attached, subject data.

The basic original data is a digital object, that has a particular meaning and can be used in a particular way. Adding metadata creates a larger complexer digital object, that can be used and operated on in more ways. (If a photo is given location metadata, it can then be used in an extra set of geographic ways.)

Metadata is a part of a data-whole that is separable, usage-wise.

Those two features – adding structure, and being separable – make the overall potential structure like a network, a graph. Each piece of data could be ‘metadata’ for another piece. You can always add something more to anything, by referencing it. Metadata is like tagging, or RDF.

This looks like what metadata is, distilled to its structural essentials. (It is not essentially about trust, or completeness, or standardisation. And it is more basic than the distinction of metadata and metacontent. (Unless this is looking too generally . . .)) It is reminiscent of Wittgenstein's ‘language games’: “a complicated network of similarities, overlapping and criss-crossing” (‘Philosophical Investigations’ s66).