RE: a question about reasoners

> From: Adrian Walker [mailto:adrianw@snet.net] 
> How would the reasoners you mention, on rdf-logic,  compare with the Internet Business Logic system ?

Unfortunately, the terms of a contract by which I am bound do not permit me to answer that question in any detail.

> There's a small example below

(elided, see [1])

Talking about parts 1 to 6 that example specifically, I worked on the GRAIL reasoner 12 years ago (best reference is probably [2], as I believe the software itself is no longer extant) that was capable of solving problems of that form.  Ian Horrocks' FaCT reasoner[3] would also have no problems with it; neither would other modern reasoners like Network Inference's Cerebra[4], Peter Patel-Schneider's DLP[5] or Volker Haarslev and Ralf Möller's Racer[6].  Fewer of those systems could, I believe, handle explanation (your part 7), principally because the developers were more concerned about clear explanations of (a) why a suggested deduction does *not* hold, and (b) chains of reasoning that may be several thousand steps in length.

All of those systems have extensive capabilities that are not exercised by that example, as I'm sure does IBL.  As several of the developers read this list (I'm sure you noted Ian slapping my wrist for some loose wording earlier :-), I'll leave them to expound further if they wish.  Sorry not to give a more detailed comparison.

		- Peter

[1] http://lists.w3.org/Archives/Public/www-rdf-logic/2003Sep/0019.html

[2] http://www.opengalen.org/

[3] http://www.cs.man.ac.uk/~horrocks/FaCT/

[4] http://www.networkinference.com/

[5] http://www-db-out.bell-labs.com/user/pfps/dlp/

[6] http://www.fh-wedel.de/~mo/racer/

Received on Thursday, 11 September 2003 05:11:24 UTC