Exploiting library catalogue data

At some point I'll catch up from when I was down with a nasty cold and do a proper non-English blog roundup installment. In the meantime this leaped out at me:

Marlen's Corner (French) quotes from a survey about catalogue use (also French) saying approximately: "[...] we must say that the quality of library data is their advantage compared to other data sources. The problem currently doesn't come from these latter, but rather from the lack of exploitation of the library data's potential by search engines, and from the lack of visibility that the interfaces give them."

Every now and then I talk about how I want a catalogue that lets users search by colour. There's just that tiny detail that we'd first need to catalogue the colour of a million-odd existing volumes and redesign the search interface... But seriously, we catalogue books with all sorts of obscure information -- by size, for example. Why do we do that? More to the point, since we do do that, why don't we exploit the fact that the information's there: why can't users search by size? Why can't we limit our searches by "has illustrations", "has colour illustrations", "includes maps"?

(Is there any catalogue that can do any of this?)