Re: PRD (Prod Rule Dialect) breakout scribe notes from F2F8 6 Nov 07

Gerd Wagner wrote:

> Dear Hassan,
> 
> 
>>>CSMA: Model theory for the operational semantics may be possible.
>>
>>Model Theory and Operational Semantics are orthogonal 
>>concepts. Model Theory fits unchanging truth. In this way, 
>>stateful computation such as that performed
>>by PR systems is at odds with Model Theory. 
> 
> 
> for PR systems we need an intertwining of model theory
> and a transition system formalism, since the state of
> a PR system can be described as a "set of facts" (this
> is the view/terminology of business rules approach),
> or more formally as a set of (finite?) models, which
> is transformed whenever a rule is fired.

Indeed - in what you propose conjugates two orthogonal
concepts: a model theory of objects with any orthogonal
operational proof theory. The former account for (immutable)
states (e.g., set of objects in WM); the latter for state
change and transitions from models to models in a family
of models. These models are vary amoung extant PR
systems, as well as the state-transition semantics, although
in both cases they may me declensed according as close
variants - indeed members of the same family of rule-based
models. This is what the contents of my slides try to explain.

>>For an example of such a formal
>>operational semantics, see a presentation I did this past 
>>summer for IFIP
>>group on Rewriting (http://rewriting.loria.fr/IFIP-WG1.6/. 
>>Slides of my
>>talk: http://koala.ilog.fr/wiki/bin/view/Main/HassanAitKaci#17.)
> 
> 
> Interesting slides. However, notice that you've simplifed
> a few things:
> 
> - Business rules are not production rules; business rules are
>   expressed by "business people" in plain English (and we have
>   to struggle formalizing them in some declarative logic, e.g.
>   in the restricted modal logic of SBVR); only after formalizing
>   a business rule you may try to implement it in the form of a
>   PR (but don't be surprised if this is not always the best
>   solution). In our case study
>   http://oxygen.informatik.tu-cottbus.de/rewerse-i1/?q=node/33#head-3
>   we show how plain English business rules can be (visually and XML)
>   formalized and then mapped to a concrete PR language (Jena and
>   JBoss/Drools).
> 
> - I cannot see how your definition of "Agenda" and the "pick"
>   operation captures what is really going on in PR engines
>   such as ILOG Jrules and JBoss/Drools. 
> 
> In fact, the challenge seems to be to define a PR engine feature 
> set that allows to define/explain the different execution semantics 
> of the major PR engines.


In that presentation, I give, not a single operational semantics,
but a family - a scheme parameterized by two arguments: the
"Agenda" and the "Pick" functions as you correctly identified.
That scheme may be dubbed PR(Agenda,Pick) and I contend that
several PR systems (including ILOG Rules) fit this scheme.
They may differ in the model theory for the objects their
rules transform, but most of them may be conjugated as a
member of this scheme as a first approximation.

As I explicated in great details in the following paper:
http://koala.ilog.fr/wiki/bin/view/Main/HassanAitKaci#14,
the frontier (the hinge so to speak) between the object model
and the state transition proof theory is formally and effectively
insured by the mere sharing or *variables* between the two worlds
of objects on one hand, and rules on the other.

What I propose needs of course to be developed and made more explicit
into a full-blown article (soon to come, I guess). At any rate, my
comments should be taken with the reserve of understanding the details
of my formal argumentation.

> -Gerd

-hak

> -------------------
> Gerd Wagner 
> http://oxygen.informatik.tu-cottbus.de/IT
> Email: G.Wagner@tu-cottbus.de

-- 
Hassan Aït-Kaci  *  ILOG, Inc. - Product Division R&D
http://koala.ilog.fr/wiki/bin/view/Main/HassanAitKaci

Received on Wednesday, 7 November 2007 15:39:59 UTC