Re: Interpretation of RDF reification

Adam -

Just to take up two of the many good points in your very thoughtful posting:

At 09:55 PM 3/26/2006 +0100, you wrote:
>Is the issue that were a more expressive language used, there would be 
>more in the user community at work on schema, data and queries utilising 
>that language?

If we try writing a small practical application as reasoning over RDF, we 
see that the proof chains quickly descend into details that are 
uninteresting and opaque to an end user.  To see this, try running the 
example called RDFQueryLangComparison1, by pointing a browser to 
reengineeringllc.com, and ask for an explanation of the answer

"Jeen Broekstra  is an author , with email jbroeks@cs.vu.nl , of  'An 
Overview of RDF Query Languages'".

Bear in mind that this is a simple academic example.

Now, one could argue that proofs and explanations are anyway not of 
interest to end users of real applications. I've heard this expressed as 
"We don't want Generals reading program traces".  A counter argument comes 
from the fact that a small change in a logical specification can have huge 
consequences, particularly if the reasoning is done over the web.  So, we 
need to be able to present human-understandable proofs at least to 
analysts, and probably sometimes also to end users.

 > Is there a particular application that would show the difference between 
the two languages
 > and prove a compelling case for would be adopters?

There are actually quite simple looking examples that cannot be computed in 
OWL.  I believe that "transitive over" is one such

(See http://www.reengineeringllc.com/demo_agents/TransitiveOver1.agent )

There are at least two issues here though.  One is, can a practical 
application be computed at all in a particular language.  Another is, will 
the process by which results are obtained be transparent and understandable 
at the end user level.  For example, compare writing the example in the paper

    www.reengineeringllc.com/Oil_Industry_Supply_Chain_by_Kowalski_and_Walker.pdf

(a) as rules in executable English, and  (b) as SQL.  Both are shown in the 
paper.

It's clear from the example that there is no hope of scaling up to writing 
a practical, maintainable, understandable application directly in 
SQL.  It's likely that the same difficulty would arise with SPARQL, given 
the similarities with SQL.

Bottom line -- for practical applications, we need sophisticated 
inferencing and explanation  tools (and IMHO also some lightweight NLP) on 
top of languages such as RDF, OWL, and SPARQL.   The "Semantic Web Layer 
Cake" can be viewed as pointing in this direction.

Hope this makes sense.  Thanks in advance for comments.

                                            -- 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 Sunday, 26 March 2006 21:54:00 UTC