- From: Michael Kifer <kifer@cs.sunysb.edu>
- Date: Mon, 27 Jun 2005 22:36:35 -0400
- To: bancroft@america.net
- Cc: Bijan Parsia <bparsia@isr.umd.edu>, www-rdf-logic@w3.org, public-sws-ig <public-sws-ig@w3.org>, www-rdf-rules@w3.org
> Michael Kifer wrote: > > >I think the (nonmonotonic) rule-based side of the stack is very mature (I > >dare say more mature than the "other" side :-) > > > Well, do you think that nonmonotonic systems (that change the > derivability relation) are preferred ? Some types of reasoning require classical logic and some require nonmonotonic logic. The primary reason for the usefulness of the latter is that it is impractical (and often impossible) to specify all the negative information explicitly. In this case, inference is made by closing off the explicit positive info. This is similar to some forms of human reasoning. > There are those that suggest that a classical monotonic reasoning system > over an instance model that changes over time only `appears` to be > nonmonotonic . . . The main reason for using default negation is not because we want the system to behave nonmonotonically. The main reason (at least in the LP and database paradigms) is what I explained above. --michael > more, > l8r, > v > > > -- > america sig > >
Received on Tuesday, 28 June 2005 02:37:25 UTC