ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] Role of definitions (Remember the poor human)

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "John F. Sowa" <sowa@xxxxxxxxxxx>
Date: Thu, 15 Feb 2007 17:51:50 -0500
Message-id: <45D4E406.9040002@xxxxxxxxxxx>
Pat,    (01)

None of the ideas I presented in this thread are ones that I claim
to have invented.  I admit I have reinterpreted older practice in
DBs and DLs in terms of newer terminology.  But my interpretation
is consistent with the definitions in the literature.    (02)

Typing "entrenchment" and "belief revision" to Google produces
14,900 hits.  Following is a survey article:    (03)

    http://plato.stanford.edu/entries/logic-belief-revision/
    Logic of Belief Revision (Stanford Encyclopedia of Philosophy)    (04)

In the that article, entrenchment is defined as a binary relation
that supports a partial ordering of statements:    (05)

    "When forced to give up previous beliefs, the epistemic agent
    should give up beliefs that have as little explanatory power
    and overall informational value as possible. As an example of
    this, in the choice between giving up beliefs in natural laws
    and beliefs in single factual statements, beliefs in the
    natural laws, that have much higher explanatory power, should
    in general be retained. This was the basic idea behind Peter
    Gärdenfors's proposal that contraction of beliefs should be
    ruled by a binary relation, epistemic entrenchment. (Gärdenfors
    1988, Gärdenfors and Makinson 1988) To say of two elements p
    and q of the belief set that 'q is more entrenched than p' means
    that q is more useful in inquiry or deliberation, or has more
    'epistemic value' than p. In belief contraction, the beliefs
    with the lowest entrenchment should be the ones that are most
    readily given up."    (06)

Another good reference is Peter G's tutorial on the subject:    (07)

    http://www.stanford.edu/class/cs224m/papers/gardenfors.pdf
    Belief Revision:  An Introduction    (08)

Following is a tutorial by Jon Doyle and Ulrich Junker from
AAAI-2004.  They use the term "preference" rather than
"entrenchment"; Benjamin Grosof uses the term "priority";
but they are compatible with Peter G's "entrenchment, and
they support a partial ordering of preferences, priorities,
or levels of entrenchment:    (09)

http://koala.ilog.fr/wiki/pub/Preference05/PreferencePortal/AAAI04_Preference_Tutorial.pdf    (010)

When Grosof was at IBM, he led a group that implemented the following:    (011)

    http://www.research.ibm.com/rules/paps/rc20836.pdf
    Courteous Logic Programming: Prioritized Conflict Handling for Rules    (012)

For the terms "entrenchment", "belief revision", and "probability",
Google has 616 hits -- among them is the first paper I cited above.    (013)

I'll admit that these papers are researchy, but the three-level
version I suggested (type hierarchy, constraints, and defaults)
includes the 30-year practice in both databases and description
logics as special cases -- as well as Cyc's 22 years.    (014)

I have no commitment to any of the three terms "entrenchment",
"preferences", or "priorities".  But I firmly believe that this
technology with at least 2 or 3 levels is mature and well tested
in practice.  Even the extensions for probabilistic reasoning
and default reasoning have been implemented and tested.    (015)

Note that I am *not* recommending any specific method of reasoning.
All I am asking for is the ability to distinguish those statements
that define the type hierarchy from other statements that define
constraints and purely factual "updates".  Other levels could be
added for defaults or probabilistic statements.    (016)

If you don't want belief revision, you can dump all the statements
in the same pot for an FOL theorem prover, but other people might
want to use them for their belief revision systems.  Providing that
option makes no commitment to any version of belief revision.    (017)

John    (018)

_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (019)

<Prev in Thread] Current Thread [Next in Thread>