Just how important are controlled subject vocabularies in the context of an IR?

Researchers submitting their work to IRs are presumably experts in their field, and  so should be fully conversant with the preferred terminology in their subject area. They can therefore be relied on to choose appropriate subject keywords to improve the findability of their work. Ditto, researchers looking for works by subject – presumably, in their own specialisation – should be fully conversant with the terminology; if they don’t find what they’re looking for under brown coal, for example, they should have a good enough sense of the literature to think of also looking under lignite. So controlled subject vocabularies are simply not relevant in IRs.

At least, so goes one argument. I’m not convinced. Brown coal might be the preferred term in the Victorian coal industry, but lignite is the term most commonly used elsewhere. Victorian coal researchers, well aware of this fact, are likely to try both terms; but coal researchers from other parts of the world may not necessarily be aware that one small part of the coal industry prefers the term brown coal.  At least, not unless they’re using a controlled vocabulary of some kind.

To maximise findability of individual works, IRs must use controlled subject vocabularies; ideally these vocabularies must be authoritative in their field – and either built into the IRs, or directly accessible from them. To assist with harvesting and sharing of metadata, the vocabulary needs to be explicitly specified in the record’s metadata.