RE: Inference over Ontology + Data

Hi Adrian, all,

Picking up the thread, my response to Adrian's query about inferencing
over the ESWC2006 data is pasted below. If other's are interested in the
rest of the discussion history it's archived at [1]. I found the
discussion held interesting examples of the kinds of questions
researchers in related fields might have about the Semantic Web.

Anyone interested in the ESWC2006 Technologies work (and the ESWC2006
Conference Ontology) referred to below, feel free to check out [2], or
drop me an email.

Cheers,

Tom.

[1] http://www.wikisym.org/pipermail/wiki-research/
[2] http://www.eswc2006.org/technologies/

<snip>
Yep, the example I gave below was simply a scenario to illustrate some
reasoning we may want to perform that could actually be useful - I
didn't expect people to actually try it with just the data I'd mentioned
;)

If you're interested in a richer data set about ESWC2006 (beyond just
the semantic delegates list mentioned in my earlier mail) you'll find
more at http://www.eswc2006.org/rdf/

However, I don't think that data set is complete enough to carry out the
reasoning in that scenario - there aren't enough "Person holdsRole
Presenter" statements (or the inverse).

Unfortunately this highlights a couple of issues: 1) that the organisers
of a conference don't always know which of the authors are going to
present a particular paper, and 2) that creating complete data of this
sort is hard, as I'm sure you know.

To solve the problem in the future we'd need some kind of lightweight
annotation tool to allow people to say "I am the presenter of that
paper". Which I guess comes back to Alain's original points about
barriers to annotation.

Anyway, lets talk more if you'd like or have more questions...
</snip>


-- 
Tom Heath
PhD Student
Knowledge Media Institute
The Open University
Walton Hall
Milton Keynes
MK7 6AA
United Kingdom

Tel: +44 (0)1908 653565
Fax: +44 (0)1908 653169
Web: http://kmi.open.ac.uk/people/tom
Email: t.heath@open.ac.uk
Jabber: t.heath%open.ac.uk@buddyspace.org 


-----Original Message-----
From: semantic-web-request@w3.org [mailto:semantic-web-request@w3.org]
On Behalf Of Adrian Walker
Sent: 18 July 2006 22:15
To: semantic-web@w3.org
Subject: Inference over Ontology + Data


Hi All --

Three of us got into a discussion that started on another list, and we
thought that some of it may be of interest to semantic-web folks.

The point of departure was some friendly but skeptical "Loaded
Questions" about semantic Wikis from Alain Desilets.  Then, Tom Heath
kindly provided a pointer to an ontology and some relevant data,
indicating some ways in which the ontology could guide inference over
the data.  So, I tried a little experiment to see if I could do this,
and came up with further questions.  

Here's an entry point to our discussion, and Tom has kindly said he will
take this up on this list, so watch this space!

--------------------------------------------------//--------------------
--------------------------

Tom  --

I was intrigued by your note to Alain, in particular where you say that

....the semantic web offers a different approach. In the ESWC2006 
ontology (which incidentally is useful for describing academic 
conferences in general - 
<http://www.eswc2006.org/technologies/ontology>) there are classes of 
things called PaperPresentation, PosterPresentation, DemoPresentation, 
etc, which are subclasses of "TalkEvents", which can be part of other 
events such as a Workshop, PosterSession, Conference etc etc. So, if 
someone has the Role of Presenter for a TalkEvent that is part of 
ESWC2006 (or part of a part of ESWC2006), we can infer that they 
presented something at ESWC2006.


So, I tried this out, and the result so far is 

   http://www.reengineeringllc.com/demo_agents/ESWC1.agent

However, when one runs this to make the kind of inference you describe,
one finds that the ontology and the data have very few (4) terms in
common, and those are not enough to support any useful kind of
ontology-guided inference that uses the data.

This could be just a matter of finding more complete data, but I'm
wondering whether there is some fundamental difficulty here too.  What
do you think?

                                     Cheers,    -- Adrian


Internet Business Logic (R)
Executable open vocabulary English 
Online at www.reengineeringllc.com 
Shared use is free

Adrian Walker
Reengineering
PO Box 1412
Bristol
CT 06011-1412 USA

Phone: USA 860 583 9677
Cell:    USA  860 830 2085
Fax:    USA  860 314 1029

Received on Wednesday, 19 July 2006 10:17:01 UTC