Panel:  What's Next?

Danny Weitzner chairing

Jon Robert Pellant of Pega:
customers want to treat rules as business assets, avoid vendor lockin, 
or reuse/repurpose 

didn't hear much about delegation model, including 
*override with more specialized behavior

need to interact with business rules from modeling tool environments

Don Chapin of BRT:

make "community" the core notion for "context" 

define a framework that goes from higher abstraction at business level 
all the way down to executable 

provide the ability to state that different expressions of terms, facts, or
rules all mean the same thing

Paul Vincent:

be compatible with other worlds, e.g, Object Oriented

TimBL: 

do something simple, 80-20
- note simplicity is more important than small-ness,
e.g., if there's a large set of things that are easy to put in, then do it

use URI's "seriously"
be "context-free"
support partial understanding

Ed Barkmeyer of NIST: 

for WG:
need an abstract syntax with one corresponding XML exchange syntax
well defined semantic model:  
 ... *thus maybe not through RDF
*must support basic datalog, constrained NAF, n-ary predicates
 ... thus not first order logic 
*must support reference to (walled inclusion of) OWL ontologies 
                                                   and other rulesets
at least one standard surface syntax,
      explicit permission for other surface syntaxes 
  - prefer frames for n-ary predicates

Kurt Godden of GM: 

interlingua that's OWL-compatible, vendor- and tool- independent 

FOL expressivity, plus a wish list:  
  lambda-binding, quantification over predicates

have Rules-Lite for domain experts 

Harold Boley: 

see his statement

Francois Abramatic of Ilog: 

we should start with simple business rules cf. IBM paper 


Discussion:

Benjamin Q:
Key to address is what will be the common technical area for the 
1. business rules community, with centroid simple business rules cf. 
Mark Linehan's presentation, and the community coming from 
commercial business rules systems
vs. 
2. the people coming from the research community (and Prolog), including KR,
and the semantic web community.  
What's the form of rules that will be useful for both of these (1.) and (2.).
I think there are 
two issues particularly critical to address, including here:  
a. negation / nonmon   in vs. out?  
    clarification:  needed for modify, retract, update, etc.  
b. actions   in vs. out? 

Jon Pellant:
negation not sure
actions yes, should be done as extension on the pure derivation part

Ed Barkmeyer:
constrained negation yes, but maybe not adequate
actions yes

Pat Hayes:
actions important, but scared about the interaction 

Ian Horrocks:
LP and OWL are really separate worlds wrt the logic.
DLP (version from original paper) is pretty weak.  

Michael Uschold:
want things driven by use cases 

Jos de Bruijn:
(missed it) 

Anthony Finkelstein:
worry about the squeeze between the large research efforts vs. the large
vendors, as a small vendor.
In addition to standard rules language, exchange form for it,
but not necessarily a single semantics and yet get interoperability advantages

Walid ___ (spelling?):  
Rule language must be simple.
Directed to the RuleML people:  ... 
I tried several times to understand the syntax and the idea, and
found it hard with so many acronyms and so many sublanguages.  
I'm sure the core is simple, but I got lost.  

Mark Linehan:
I think Benjamin's first point about what's the thing that is common
between the business rules and is very important, coming at this as a vendor.
I don't think it's clear yet.
I'd like to see it described in the kind of way, and simplicity, that
the OMG PRR is described.

TimBL:  no one is suggesting that rules be exchanged directly as RDF graphs.

Pat:  
- from recent DoD event:
    wants unique name assumption, that's a key nonmon thing they want
- it's no brainer that we want n-ary 

___:  want uncertainty

Bijan:
there are some important issues that use cases don't solve, 
including compatibility and reconciliation of communities, theoretical, ...

Gary Liu:  there's still a communication problem between the business community
and the KR community, and a split in KR community too

would be nice to create a table of semantics and features and what they
enable, that would help users and vendors to pick and choose features

____:  does the business community understand what they want for 
Rules Lite?  vs. Rules-Full

Murray Burke:
it's important to think in terms of tasks, e.g., 
policies etc. as Benjamin talked about; 
semantic web services;
note there are non-business uses for sharing of rules

Adrian Walker:
glaring omissions:
1. when running rules distributed, things get complicated fast,
and need to have ability to extract explanations 
2. Google-friendly form of rules

Mark Linehan:
It's OK to start small.
Idea: maybe it would be fruitful to work on even smaller building blocks,
e.g., using URI's or building arithmetic expressions

Windup remarks by panelists follow

Don Chapin:

SBVR incl. its use cases might help with finding the subset that 
is common between the business rules and KR/SW folks

Paul Vincent:

broad church.
good news is that I haven't seen any showstoppers

Jon Pellant:

agree with Paul.

TimBL:

wrt that gap, we've made better connections and we're making progress,
there's a clearing in the woods

Ed Barkmeyer:

we should avoid danger of spending a year in a WG fighting over charter,
be clear going in what the technical scope is.

OMG lot and RuleML lot will soldier on even without us.
Would be good to avoid competing standards. 

Kurt Godden:

I think there's incredible potential to improve business here.

Harold Boley:

would be good to compare existing builtins with the OMG folks.

would be good to try to extract semantics from SBVR and similar.

Francois Abramatic:

We have to find a way to get the momentum going.