Re: Atom and RDF

On Tue, 12 Oct 2004 07:16:29 -0700 (PDT), Dare Obasanjo <kpako@yahoo.com> wrote:
> 
> --- Danny Ayers <danny.ayers@gmail.com> wrote:
> What is this N^2 interop problem and can you explain
> how it doesn't apply to using RDF? I've read Ken
> MacLeod's postings on the topic but as an aggregator
> author I felt his arguments were specious.

I thought Ken put it rather well, but it's easy enough to describe in
terms of aggregators. An aggregator with RDF support can aggregate
arbitrary data (say RSS plus the author's personal information and
their work schedules) as long as it's expressed as RDF/XML. It can
store it. Ok, you could argue that all that could be done with XML
alone. But the big difference is that because the source data has been
expressed as RDF it has consistent semantics that can be used for
query or inference.

So I could run a query like "give me all the items friends of Joshua
posted on days when he was on vacation". You could build a similar
system based on XML or a RDBMS. But the N^2 problem kicks in when you
try to add a new vocabulary or terms - say "starsigns". Exactly the
same store and query mechanism can be used to ask "give me all the
items people with the same starsign as Joshua posted on days when he
was on vacation".

Every term/vocab you add, has to be fitted it in with the existing
infrastructure, i.e. the N systems you've already got. Whereas if your
original data is expressed in RDF, you only have to fit in with 1
system.

This doesn't work for XML alone as there isn't any consistent
interpretation above the syntax. RDF can do it because it maps to
certain bits of pretty well researched bits of predicate logic. It
really justs pushes the shared language up a level is all.

Cheers,
Danny.

-- 

http://dannyayers.com

Received on Tuesday, 12 October 2004 15:48:46 UTC