[Pellet-users] Reasoning Efficiency in Pellet

[Bijan Parsia]

On Sep 4, 2007, at 5:55 PM, Alan Ruttenberg wrote:

> Is your only criterion efficiency? In my work, which deals with  
> real world entities, I tend to make the choice of classes versus  
> instance based on whether the things I am instances, which are  
> chosen to be traceable to specific things in the world, and  
> classes, which are collections of those things.

Good point, and I'll add that instances vs. class can be somewhat  
application/interest dependent (just as atomic objects vs. composite  
ones can be).

After all, you can model a person as an individual (perhaps with  
parts) or as a set of time slices.

> While this does mean that efficiency can be a problem sometimes,  
> the representation is less likely to be have to change based on  
> which reasoner you use, or a new advance in reasoning technology.
>
> OTOH, if you are just trying to implement some calculation, such as  
> with the Soduko example of Bijan's, then perhaps efficiency is the  
> only thing that matters.
>
> I'd be glad to chat about this with you, if you are interested,   
> but I'm not sure it is a matter for the Pellet list. 'Course you  
> could reply here and wait and see whether Bijan kicks us off :)

It doesn't seem unreasonable at the moment, but you might find a  
broader audience and a wider range of participants on a more general  
purpose list. Whether this is something to be sought after or avoided  
I leave to you to decide :)

Cheers,
Bijan.