Re: isPartOf inferencing, was Longwell display of RDF...

From: Ryan Lee <>
Date: Mon, 16 May 2005 11:14:22 -0400

Jon Crump wrote:
> On Thu, 12 May 2005, Ryan Lee wrote:
>> Inferencing can be a very powerful addition to your toolkit for
>> handling RDF. I am personally most familiar with cwm.
> I was under the impression that longwell performed some kind of
> inferencing; if not this kind then what? (It's particularly interesting
> to me that when I let longwell loose on a set of rdf files, it tells me
> there are x number of statements in the resultant model before
> inferencing and x+ statements after. It would be extremely instructive
> if I could see what those additional statements were).

The inferencing we do is limited to deducing subclass relationships and
making deductions that follow from owl:sameAs relationships, but nothing
else. We've found the inferencing code available to Longwell 1.x to be
too inefficient to be useful beyond what it now does, and even that can
be a long wait.

> If I constructed some separate ontology in skos or owl or something like
> that that defined rules for such implications or entailment, would
> longwell then be able to display them in the facets I've established, or
> do I need a completely separate tool (cwm?) to explore and display these
> kinds of relationships?

You want to display the set of results from running your rules? If the
conclusions of your rules are just more RDF in the same 'shape' as what
you're presently putting into Longwell, there shouldn't be a problem
with that.

Cwm has a particular rule syntax, so you wouldn't specifically be using
OWL or SKOS to define your rules.

> Can cwm aggregate rdf models, make inferences, enforce entailment and
> then produce a new composite model out the other end which longwell
> could then browse?


Contrary to its name, cwm operates on an open world assumption. While
it will aggregate RDF models and make inferences based on rules you give
it, enforcing entailment may not end up meaning what you think it means
if you aren't used to open world operations. For instance, a correct
conclusion to derive from the following in an open world:

:this :is :true .
:this :is :false .

would be that

:true = :false .

so cwm will have no trouble concluding the preceding statements when
generating its final output. You could instead write rules to flag an
error if something you know is unwanted is generated.

More resources on where to go for asking about cwm:

And a tutorial about writing rules here:

Ryan Lee       
W3C Research Engineer    +1.617.253.5327
Received on Mon May 16 2005 - 15:13:17 EDT

This archive was generated by hypermail 2.3.0 : Thu Aug 09 2012 - 16:39:18 EDT