Re: Intersection of properties?

>> It seems that people have taken these approaches:
>> 1) Only use small toy ontologies like "pizza",
>> 2) Use a large ontology but with only very simple structure such  
>> as names and addresses,
>> 3) Rely on Pellet to "perform a consistency check before it starts  
>> a reasoning task" without knowing how long it will take or whether  
>> the algorithm it uses is valid.

A couple of comments:

A consistency task *is* a reasoning task. It can be shown that the  
other common reasoning tasks for OWL can be reduced to consistency  

There is another option, which is the approach we have taken with the  

4) Build large ontologies that use as much of the expressiveness of  
OWL as we need to. Do sound and complete reasoning on portions of it  
using Pellet or Fact++ as sanity check or for specific tasks. Do  
incomplete reasoning on the rest of it to get possibly incomplete  
results on particular tasks using whatever strategies we can.  
Challenge the theorists and reasoner developers to do more with these  
artifacts (they seem to like it when they are given real, well  
modeled, challenges). Among recent developments in the direction of  
reasoning developments I can point to the release of SHER, the  
development of Hermit, and the implementations  (current and  
upcoming) of the tractable fragments Michael alludes to.

> Only if no one of the internally known languages match, then an  
> original OWL-DL reasoner will be used as a default reasoning  
> engine, which may or may not lead to poor efficiency.
> I hope other people in this list can confirm my guess here, and can  
> elaborate on the state of the art a bit.

It's my understanding that this is already the case with current OWL- 
DL reasoners. For instance one issue I hit a while ago was that the  
mere presence of an inverse property statement (but no use of the  
property) caused an ontology I had to stop being able to be  
classified in available memory/time. The reason, it turned out, was  
that Pellet was deciding which reasoning strategy (implemented by  
different code) to use, and the one that understood inverse  
properties did worse on my ontology.

I don't know whether the simple check for the presence of the  
definition is still all that is done, but one can see obvious  
modifications that would improve this, for example checking that the  
inverse property is only used in simple facts and in that case  
rewriting them backwards to avoid the use of the inverse.


Received on Wednesday, 6 August 2008 16:03:39 UTC