Re: Rules WG -- draft charter

Hi folks,

Wrt the role of justifications in the charter of the proposed Rules WG effort:

I very much agree with Ian's viewpoint (below).
I was in the midst of drafting a message making much the same points when I 
received his.
So I'm just including here my additional points.

Justifications are certainly important.  The area of justifications for 
semantic web rules (or ontologies) is however not very mature,
it's still pretty research-y.   I could get into an explanation of how 
nonmonotonic logic program proofs are not much more computationally complex 
in time and space than monotonic/Horn logic program proofs, but that's not 
really the point as far the charter agenda discussion we're having now is 
concerned.  Indeed, the fact that that explanation needs to be given at all 
to so many folks involved in semantic web rules just illustrates that the 
area of justifications is indeed research-y.

Justifications is mainly separable from the rules language 
itself.  Division of labor is more effective.  A W3C WG requires a lot of 
labor resources from its participants and aims for a fairly predictable 
timeline  -- 1 year or less to fairly technically mature standards 
recommendation.  Thus as a strong principle a W3C WG should be based on 
relatively mature research foundations.

This will increase the WG's probability of success, reduce the overall 
amount of labor it takes, and reduce overall elapsed time to reach its 
result.  It will also make for a less burdensome and happier time for most 
WG participants, increasing the chance they / their organizations / others 
like them  will participate in future W3C semantic web standards efforts.

What I suggest for the Rules WG Charter is that we mention as a requirement 
for Rules that the rules language should be able to support a 
future-developed justifications language standard, and pretty much leave it 
at that.

The Semantic Web stack diagram shows proof and trust (a bit vaguely) as 
later-developed layers that depend upon the rules (and ontology) language, 
and that's the right overall approach at this point in history.

In short, we should greatly deemphasize justifications in the Rules WG 
Charter relative to its current draft (V1.26).

I have to say that of the folks I have talked to since Friday who have been 
heavily involved in the main early-phase semantic web rules standards 
efforts (RuleML and Joint Committee), most were quite surprised by how 
large the role justifications play in the current Rules WG charter draft 
(V1.26).  That should tell us something, too.  It wasn't on the primary 
agenda for a Rules WG before, and I think there are compelling reasons to 
keep it off.

Benjamin

At 01:27 PM 11/11/2003 +0000, Ian Horrocks wrote:

>On November 7, Sandro Hawke writes:
> >
> >
> > As Dan Brickley mentioned, we've been working on a Rules charter, too.
> > Here it is:
> >
> >        http://www.w3.org/2003/10/swre578
> >        (currently at revision 1.24)
> >
> > It's still rough in places, but I think it gets the point across.
> > Wording suggestions are welcome, as are questions about what is meant
> > by some section or phrase.
>
>Executive summary: rules and justification are at completely
>different stages of development, and linking the two threatens to
>sabotage any rules WG - please don't do it!
>
>
>As is clear from the draft, rules are relatively well understood
>theoretically, and there are already several proposals on the
>table. The WG would, therefore, only (!) have to settle issues such as
>expressive power and syntax (ever the optimist).
>
>As is also pretty clear from the draft, the situation
>w.r.t. "justification languages" is completely different - in contrast
>to the detailed sections on rule expressiveness and syntax, there are
>only a couple of vague sentences about justification. It is not the
>case that "justification" is well understood theoretically (it is
>still very much an open research problem) and it is far from clear
>what form justifications would take (this might be heavily dependent
>on the logical language as well as on the reasoning technology being
>employed), never mind what language should be used for expressing
>them.
>
>If, as is stated in the charter, "The readiness and urgency in this
>field are growing" and "The full benefit of the technology really
>depends on the suite of standards being complete", then it seems crazy
>to sabotage the rules working group by asking it to additionally solve
>the problem of justification and justification languages.  Therefore,
>in the interests of making rapid progress on rules, I would suggest
>that the two tasks are NOT linked, and that a separate
>interest/working group is chartered to address justification.
>
>Regards,
>
>Ian

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

Received on Tuesday, 11 November 2003 09:52:24 UTC