Re: "tdb" and "duri" URI schemes...

Graham Klyne wrote:
> Nathan wrote:
>> FWIW, duri scheme appears perfectly clear, tdb with fragments or where 
>> a doc describes multiple things (_Things_ described by) isn't so clear.
> 
> As I understand it, and given the typical handling of URI#frag using 
> RDF, tdb: with fragments is probably redundant.
> 
> But there have been situations where some way of using URIs without 
> fragids in similar fashion was desired, which is where I think tdb: 
> comes in to play.  To riff on your earlier example (and adopting 
> Jonathan's suggestion to drop the date in tdb, which I liked):
> 
> [[
>   <http://webr3.org/nathan> a document primarily about me
>   <http://webr3.org/nathan#me> me
> ]]
> 
>   <tdb:http://webr3.org/nathan>
> 
> could be another way of referring to the same thing as
> 
>   <http://webr3.org/nathan#me>
> 
> The ideas in Larry's draft have been kicking around, and periodically 
> resurfacing, for a long time now and I think it's about time to move 
> them to a stable and citable form (at least experimental) so we can see 
> if they find any traction in the wider web.

This issue seems to be at coming to the fore at the minute on the 
semantic-web list, will be replying there in a moment, however I do 
kind-of agree, and would think that making the date part of tdb: 
optionally would boost adoption, however it still doesn't solve the 
issue (afaic) because, document to thing described is not a 1 to 1 
mapping, many documents describe multiple things, so the need is for 
"things described by" and not "thing described by" to cover all use cases.

However, way deep down at the heart of all of this is that we need to 
ensure that different things have different names, so that when yuo 
consider the giant global graph, I am not a document with a media type 
and so that a document isn't a Person with a gender and blood type. The 
two exiting patterns of [ /doc and /doc#thing(s) ] and [ 303 See Other ] 
to some degree address this, and both have negative impacts on web arch 
in some way. And likewise tdb: addresses this but has negative impacts 
in other ways.

I do however like the temporal aspect which duri: and tdb: provide.

The critical bit though, and the one we need to ensure is preserved, is 
that different things have different names, so that in 2020 when 
somebody downloads the RDF Graph as of 2010 in a *TB sized zip, they can 
actually use the data and reason over the statements, successfully.

Personally, and for many reasons, I've chosen the [/doc and 
/doc#thing(s) ] pattern, it allows me to be network friendly and 
describe multiple things in a single document, whilst ensuring that in 
all contexts my things are never also identified as being documents. I'd 
also very much like to introduce a temporal aspect, by way of duri:.

As for tdb: well I can't see me needing it with my own things, however I 
can see me using it when linking to all the URIs which use the 303 
pattern :)

Thus, I hope that duri and tdb allow the following usage:

   <duri:2010:http://webr3.org/nathan#me> :interested_in 
<tdb:2010:http://dbpedia.org/resource/Resource_Description_Framework> .

Best,

Nathan

Received on Thursday, 4 November 2010 17:25:03 UTC