Begin forwarded message:
> From: Jonathan Borden <>
> Date: June 9, 2004 9:01:16 PM EDT
> To: Elliotte Rusty Harold <>
> Subject: Re: [xml-dev] Meta-somethingorother (was the semantic web
> mega-permathread thing)
>
>
> Elliotte Rusty Harold wrote:
>>
>>
>> I'd like to see the use cases where starting from an RDF triple store
>> makes life easier than starting from the identical information in
>> more traditional XML. Use of all easily available free tools and
>> languages is allowed (Python, Java, XSLT, XQuery, Owl, etc.) Reliance
>> on expensive proprietary systems (e.g. Tamino) is not because as I
>> saw on sign somewhere recently, "If we can't afford the solution,
>> it's not a solution."
>>
>
> There is no inherent advantage to RDF triples *except* that it is alot
> easier to canonicalize triples in, e.g. the N-Triples format than it
> is to canonicalize XML, or any particular dialect of XML. All this is
> saying is that if you have data that let's say is naturally
> represented as a directed labelled graph (DLG) and for which you might
> have 1000 slightly different ways to represent that in XML, triples
> *might be* a least common denominator format to represent the data.
> Might be.
>
> For example RDF/XML. The folks on the RDF Core and OWL working groups
> soon found that it was somewhat difficult to communicate using RDF/XML
> (Lots of these discussions were naturally about triples). It turns out
> that it is alot easier to just *say* the triples rather than try to
> express them in RDF/XML. N-triples (and TimBL's N3) is a convenient
> shorthand to author information naturally expressed as triples.
>
> For actual use cases which employ free tools, take a look at the N3
> work that TimBL and DanC have done on N3 and the CWM inferencing
> engine all of which is written in Python. The strong impression I get
> is that for alot of folks who are hacking in RDF, the RDF/XML syntax
> presents a big overhead.
>
> For example here's a little N3 thinging to do with Atom:
>
> http://semtext.org/atom/atom-n3.php
>
> which "starts" as an RDF triple store. One could express the same
> definition in RDF/XML but I understand that a number of folks who are
> doing this stuff find it easier to work with N3 *for this stuff* than
> RDF/XML -- and that would be the identical information.
>
> Jonathan
>
