If you want an ambiguous source of information, use natural language. The power of data is that is a whole lot less ambiguous.

TimBL, entering a pretty heated debate going on right now, stemming from the recent schema.org announcements from Google, et al., essentially concerning the level of semantic precision that is necessary to build the semantic web. There was a tweet the other day about the speed at which TimBL both speaks and writes and this response suffers a little in its haste I think, but it captures the essence of the argument quite nicely.

Linked Data purists argue that if you want to describe the Eiffel Tower and a web page about the Eiffel Tower then you have to have a way of identifying them separately (and unambiguously) - otherwise you run the risk of describing how tall a web page is :-)

Others argue that some fuzziness is acceptable, for pragmatic reasons, and that systems will be able to use context to determine which descriptions refer to what.

I noted on Twitter the other day that this is a debate that has been going round and round for at least the last 10 years to my knowledge - and I've been more and less involved at various points. It was certainly something that we used to discuss a lot during Dublin Core architecture meetings.

I don't see the debate stopping, or being resolved, any time soon to be honest.

EIDR is a universal unique identifier system for movie and television assets. From top level titles, edits, and DVDs, to encodings, clips and mash-ups, EIDR provides global unique identifiers for the entire range of audiovisual object types that are relevant to entertainment commerce.

(Sorry... not particularly new but I just found it).

Chalk another one up to the relentless march of the DOI.