Re: New Intro to RDF

Thanks for everyone's comments so far.  I'll respond to all the comments 
  in one shot:

Hammond, Tony wrote:
> One thing you may want to change, though, throughout is your use of 
> 'urn:' URIs. The '/' character is specifically reserved in 'urn:' 
> syntax

Okay, good point.  My goal was to point out that URIs aren't necessarily
confusable with web stuff, but I guess I ended up with something that
was a valid URI on the face of it but an invalid URN.  What would be
nice is a URI scheme that has the same hierarchical syntax as http, but
without bearing any similarity to a web-oriented URL.  Maybe an 'id'
scheme to make hierarchical identifiers.

tim.glover@bt.com wrote:
> Under the section "What is RDF" you say "RDF is nothing more than a 
> general method to decompose information into pieces".
(snip)
> On the one hand, RDF is commonly used and treated as a database of 
> information held in "triples".
(snip)
> On the other hand, actually RDF IS more than this. It has a formal 
> logic which allows you to deduce...

I didn't mean to split those two views of RDF.  Really I meant to split
those two together from the RDF-as-XML view.  What I should say in the
article is that the 'general method' I refer to comes along with some
semantics as well.  But, importantly, almost all of the semantics used
for deduction is introduced above the level of RDF, with RDFS and OWL
and other schemas people come up with.  So looking strictly at RDF, the
semantics provided for deduction isn't very interesting.  I will add
more about the semantics to the introduction of the article, though.

> Secondly, I would like to make a comment about the use of URLs as 
> unique identifiers.

Aside: There was a nice back-and-forth in some blog post + comments
linked from planetrdf not too long ago.  That was a good read...

> Names only have to be unique IN CONTEXT.

I don't buy this argument.  Let's say we take away the rule about URIs
being globally unique from the core of RDF.  Now I'm given two RDF
documents.  In fact, let those two RDF documents be identical.  I can no
longer tell if the documents have the same meaning.  If in the general
case URIs don't need to be unique, there is no way within RDF to
indicate that two identical URIs in separate documents are supposed to
be treated as meaning the same thing, so effectively they never do.  The
whole point of using RDF for distributed information disappears.  In
RDF, URIs must be unique for RDF to have any use.

That's not to say that in practice URIs must absolutely always be
unique.  One can layer context-sensitive semantics on top of RDF to
handle context-sensitive URIs.  But RDF alone is useless without an
assumption about global URIs.

> Using URLs makes many people believe that the URL is an address of 
> some useful information

I'm explicit about this in the article that URIs aren't to be peered
inside of.  And, when lousy URIs are chosen for resources, that's not a
problem with RDF per se, that's a problem with people choosing a lousy
URI scheme.  (IMO)

> URLS make RDF difficult for humans to read and understand. The 
> problem is compounded by that fact that prefixes can be used in some
>  places but not in others.

I don't find URLs difficult myself, especially when they're written with
prefixes.  And the problem with prefixes being used sometimes here and
not there is just a problem with particular serialization formats, which
is exactly what I'm trying to say RDF isn't.  (I agree with you about 
that problem in RDF/XML, but I'm just saying it's not a problem for 
RDF-the-method.)

editor@content-wire.com wrote:
> One question I have come across lately is comparison of rdf vs opml, 
> as the latter is purported in my understanding to having a similar
 > function

I haven't studied OPML much, but my understanding is that OPML has 
really no similarity to RDF.  OPML is a specific XML serialization 
format for representing a hierarchy (::cough:: I thought that's what XML 
was for to begin with), without any specification of the 
semantics/meaning of the hierarchy.  RDF is, on the other hand, a method 
  for representing any sort of knowledge/meaning, with many recommended 
serialization formats beyond XML.

Thanks again to everyone who responded.  I hope to improve the article 
based on the feedback later this week.

-- 
- Joshua Tauberer

http://taubz.for.net

** Nothing Unreal Exists **

Received on Tuesday, 4 October 2005 12:52:57 UTC