I find I use the term quite a bit in my writing, but it doesn't really have a definition. Here's what I mean by it:
Robust data is data that is constructed to survive and function in multiple settings. It's reusable. It can be updated. It comes with links to all its sources. Its provenance is explicit and transparent. If you ask a question about it, everything known about a piece of robust data produces an authoritative answer -- including answers about its limitations. Robust data is portable, packable, and machined to precise standards. It's published in a standardized schema, can be retrieved precisely and there are no questions of ownership when you buy it and sell it.
Robust data has a single definitive representation in a system of knowledge, which means it has a URI within a curated directory.
Recent Comments