my non-scribe notes from the classification breakout at the kickoff today

Hi Chris -- and Pascal,
You asked I send my (non-scribe) breakout notes to you.  They are below.
I'll plan later to edit/contribute to the RifClassification wiki as well.

I figure it doesn't hurt to post these notes to the mailing list as 
well, so I've cc'd that.

Benjamin

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
*non-scribe* notes from the Classification breakout at the RIF 
kickoff  12/10/05
by Benjamin Grosof

bases for classification of rule systems/languages:
- market segments and coherence / history
- expressiveness

KR expressiveness:
- declarative LP
- FOL
- modal:  deontic; temporal

dimensions also include:
- nonmon, defaults, priorities
- procedural attachments

also there's:
- uncertainty/probability

families:
RDBMS (relational database management system) core SQL
- XQuery and SPARQL as outgrowths of that
PR (production rules)
ECA (event-condition-action rules)
Prolog
also:  sequential rules, related to PR especially

also:
FOL -- for ontological etc. modeling, e.g. cf. SWSI effort
OWL -- could think of OWL as a rule system

unique names

there are control constructs in the language in for example PR and Prolog


Benj and several:
think in terms of expressive features and restrictions

see Benj's slides from kickoff presentation yesterday,
esp. slides 41-43 there;
it's on Benj's webpage -> Talks

Pascal and Benj:
can aim to have an OWL ontology of the features and restrictions

tasks:
inferencing -- this is primary

other tasks:
authoring and modeling
analysis:  used to support inferencing or authoring/modeling
also:  testing:  uses inferencing, analysis

control of inferencing:
- backward
- forward
- mixed
- incremental vs. exhaustive

operational

wrt modeling-level info:  important here are:
transformations and round-tripping, passing along metadata,
provenance, commenting

secondary but significant concern is support not just of inferencing
but authoring/modeling/analysis

Michael K:
we should classify in terms of syntax and semantics
syntax includes:  plain predicate calculus, Frame syntax, reification, etc.
types of semantics wrt mon/nonmon: open world, closed world well-founded,
    closed-world answer set semantics

discriminator attributes between systems:

FOL-minus, FOL, FOL+
open vs. closed
procedural vs. declarative semantics
uncertainty
modality, intensionality
computational complexity/decidability

look at past standards design / standards work:
RuleML and SWSL
SWRL
CommonLogic

let's build up a wiki
RifClassification
- Pascal will build it up

(Chris W. had to leave at this point)

Benj:  expressively, let's also work from supply side, bottom-up:
what we know how to do interoperability-wise, pairwise, then
take a superlanguage of the expressiveness of those pairs (pairwise bridges)

SBVR took somewhat restricted CL:  FOL + Henken semantics
then added alethic and deontic:  necessity and obligation
key people:  Pat Hayes and Terry Halpern

Donald Chapin:
can we capture in RIF what's in SBVR
- see Annex A of SBVR standard

TODO:  all to send notes to Chris Welty who will aggregate


________________________________________________________________________________________________
Prof. Benjamin Grosof
Web Technologies for E-Commerce, Semantic Rules, Business Policies, 
E-Contracting, Services, Trust, Financial
MIT Sloan School of Management, Information Technology group
http://ebusiness.mit.edu/bgrosof or http://www.mit.edu/~bgrosof

Received on Saturday, 10 December 2005 00:12:04 UTC