Lists Home |
Date Index |
Michael Rys wrote:
> Once somebody shows me a description logic inference engine that
> actually scales, I will become more interested in this aspect.
I suspect some people are gambling on the idea that you parallelize
the engines and aggregate the results, instead of building very big
ones. You start thinking in terms of a distributed search problem
insted of a trying to scale a query/inference engine to deal with
very large datasets. But it depends on what you mean by scale.
Theoretically I guess you get some "webby" leverage by being able
ask questions of the all the DL capable nodes in a uniform way the
way you can ask any webserver GET.
> The one advantage RDF as any extended entity relationship model has over
> tree models like XML, is the ability to represent relationship graphs.
> However, for that, there is a very well suited tuple based model: the
> relational model.
Indeed. I don't see that the RDF model has significant advantages
over the relational one (that RDF uses globally unique keys is
something of a detail). But, for my uses I find being able to gather
up triples from various points in a distributed system to do things
like track messages very useful and very lightweight and less
working than having all the points emit or agree to XML formats or
to have a standard stack in place. I guess what I do with RDF here
is like data warehousing for systems administrators or an
application level syslog. If I formalized things I might/could just
use an XML grammar - we talk about this in work from time to time
and seem to verring towards RSS/Atom as packaging format for such
triples, but the point is that triples seem to be very handy to
By the way, I don't use RDF/OWL for doing like ontology work or
defining document formats.