Re: Preparing for metadata architecture discussion at the F2F

On 10/15/2010 3:05 PM, Jonathan Rees wrote:
> On Fri, Oct 15, 2010 at 1:28 PM, Noah Mendelsohn<>  wrote:
>> A few n6tes 6n the attached.
>>> 1. sadly, representations and metadata subjects do not generally have
>>>      their own URIs,
>> On purely architectural grounds, this always seemed a strange asymmetry to
>> me, given that the general Web philosophy us: identify everything of
>> interest with URIs.  I understand why in practice this might typically be
>> overkill, but it's interesting to see it emerge as a shortcoming here too.
> Of course you can use a URI to name whatever you want. I was not
> making an architectural statement, just making an observation of fact.

Right. What I was lamenting is that there is not in general a way to get 
URIs for every representation.  I think it would have been architecturally 
interesting, if not necessarily a cost/benefit win in practice, to have 
such a requirement or capability in the architecture.  We agree that you 
can create a URI for any particular representation, but there is no 
standardized means of doing so.

> Many http: URIs make perfectly good names for metadata subjects, as
> there is credible institutional commitment to "representation"
> stability. But these are the minority among http: URIs in the wild.


>>> However, any metadata assertion (author, title, etc.) stated using a
>>> URI should be approached with caution, as the metadata subject you
>>> would see now might not be the one to which that metadata originally
>>> applied.
>> A few years ago I toyed with the thought there might be some way of
>> explicitly indicating, probably in HTTP headers, representations that were
>> guaranteed to be invariant for all time, in the sense that subsequent
>> retrievals would in some well-specified ways be "the same" (though clearly
>> not the same in all properties, such as "time of last retrieval").  Anyway,
>> the idea of offering such an HTTP header seemed to land with a pretty big
>> thud, so I won't pursue it unless there is new interest on the part of
>> others.  I do think it makes the Web a bit more rigorously applicable in
>> situations like this.
> TimBL's genont ontology does something close to this, and the metadata
> could easily be deployed using .well-known/host-meta and/or Link: -
> deploying such metadata was exactly the subject of my blog post. The
> question is whether there is a market. If there were, the interested
> parties could without much difficulty get together and coordinate on a
> standard, or a first mover could just pick a solution they liked. I
> just don't think there's that much interest at present - the
> information could be provided, but who would use it?
> Jonathan

Received on Sunday, 17 October 2010 01:07:26 UTC