Re: tag: and urn:tag: considered unnecessary

> I am writing to suggest that the tag: and urn:tag: schemes proposed in
> http://www.ietf.org/internet-drafts/draft-kindberg-tag-uri-04.txt are
> not necessary, being already subsumed in the existing urn:newsml scheme.
> 
> urn:newsml has an unfortunately over-specific name, I admit from
> the beginning.  But it has essentially the same structure as tag: and
> provides the same benefits.  A newsml scheme URI looks like this:
> 
> 	urn:newsml:whatever.com:20010911:anyoldstring:1
> 
> This is, semantically speaking, a tag minted by whoever held
> "whatever.com" on 2001-09-11, and is named "anyoldstring".  The 1 part is
> a positive integer representing a version number: this is not optional,
> but can always be set to 1 for people who don't need versioning.

There is a lot of overlap.  This should be pointed out in the tag FAQ
or the draft.

The biggest practical difference is that newsml requires minters to
own a domain name.  This may be reasonable for publishers of news
articles, but it's an unnecessary restriction in general, so the tag
specification also allows owners of e-mail addresses to mint tags.

Semantically of course there's a problem that RFC 3085 says urn:newsml
resources are "NewsItems".  The idea is to name lots of things (cars,
stars, bars, guitars, ...)  with tags.  Most software will never
understand or care that these things are not NewsItems, but ontology
systems (see [1]) certainly will.

3085 also says that urn:newsml URIs map 1-1 to their resources; this
is not appropriate for many tag: applications.  There is no way to
stop people from naming things, in general, even if they already have
names.  Systems with unique names for things are nice, but impossible
to construct in some situations.  

It seems to me, without really knowing anything about NewsML, like it
wants to be using tags.  I imagine the developers had to make a new
URN NID to go with their markup language because they didn't know
about tags, or because tags hadn't made it through the IETF process
yet.  I'd be interested to hear from a NewsML creator on this
question. 

    -- sandro                         http://www.w3.org/People/Sandro

[1] http://www.w3.org/2001/sw/WebOnt/

Received on Friday, 21 March 2003 16:45:38 UTC