On Aug 1, 2014 12:35 PM, "[log in to unmask]" <[log in to unmask]> wrote:
> This is very wise. I'll add two cents of my own:
> 1) Many (not all, but many) of the uses of controlled data in systems like RDA will be (better) supported by the use of identifiers in Linked Data.
> 2) Working on the Web, it's likely that many of the uses of transcription will be better supported by other forms of representation, like digital reproductions of a resource (e.g. scanned images) or even complete access to a (digital) resource.
> We ought, I think, to take these factors into account when deciding how to invest time and effort into supporting these two forms of description.
Heh. I'll chime in with thanks for openly suggesting the radical approach of digitizing the transcription. It's something I've been grinding my teeth over and have occasionally thought "wouldn't it be nice if..." for quite some time now.
There's still a need for an accessible version of that information, of course, but I would be so much happier if we we first and foremost identified the manufacturer/publisher/distributor and their location with identifiers (in a MARC world, I suppose that would mean hammering out 7xx fields with appropriate relator codes) rather than focusing on the transcription statements with all of their abstruse punctuation.
As it happens, I was just working on trying to represent RDA 264 fields in Evergreen using schema.org last night, and found frustratingly little I could salvage without unleashing a separate NER process. Check out this commit message for a possible laugh: http://goo.gl/aWuipC