W3C home > Mailing lists > Public > public-rif-comments@w3.org > July 2009

comments on RIF: important, needed, next steps

From: Benjamin Grosof <BenjaminG@vulcan.com>
Date: Fri, 31 Jul 2009 15:01:07 -0700
To: "'public-rif-comments@w3.org'" <public-rif-comments@w3.org>
Message-ID: <CC28F43ED4708D489ABCF68D06D7F5560300DF17DD@505DENALI.corp.vnw.com>
Hi folks,

Some comments on RIF are below.

The current specification(s) and associated documents are a good job overall.
Let's keep up the steam in the home stretch.

More salient is that this initial step of RIF BLD/Core/PRD/FLD is important.  It is needed by industry, research, and the web overall.
It is a key step on the road to reduce fragmentation in the business rule industry and semantic web/technology industry, and to have a number of benefits for customers.
See [3] for an industry roadmap analysis I did a couple years ago on this topic.

After RIF becomes a Recommendation, there is more work to be done under the extensibility framework (FLD) especially.

PRD is a good initial step but essentially has a procedural operational semantics rather than declarative model-theoretic semantics, thus is
not "fully" semantic in that sense and does not support interoperability nearly as satisfactorily as logic dialects/approaches that do have such.

Notably needed in extensions (i.e., new logic dialects) under FLD are two kinds of expressive features:

1. nonmonotonicity -- default negation, and then more expressive defaults.

2. actions -- conclusion-triggered external procedurally attached actions, and then events too, similar to those in production rules and Event-Condition-Action rules.

In terms of "80-20" kinds of thinking, "80" percent of current and potential
commercial applications in the business rules and semantic technology sectors need nonmonotonicity and/or actions.

These two features were identified as high priority at the time the Working Group was being chartered, but have not yet been incorporated into the Logic Dialects
that are detailed in the current RIF specifications.

Much of the basis for such extensions under FLD is available in standardization-oriented KR semantic rule designs such as RuleML [1] and SILK [2].
These do have fully declarative, model-theoretic semantics.

Indeed, the SILK effort by the group I lead at Vulcan is currently developing some such extensions,
with contractor partners that include BBN Technologies, Stony Brook University, and others.

All that awaits is to get the current RIF stuff, done by the Working Group, out as an actual Recommendation.

I think we should all be looking forward to that.
Relative quiet on the public comments etc. mailing lists mainly indicates lack of controversy, in my view.

Sincerely,
   --Benjamin

Benjamin Grosof, PhD -- Semantic Technologies.
http://www.mit.edu/~bgrosof/

Sr. Research Program Manager, Vulcan Inc.
Head of Project Halo Advanced Research (HalAR) program.
http://silk.semwebcentral.org
http://www.vulcan.com

and

Founder and Principal Innovation Consultant,
Benjamin Grosof & Associates, Inc.
http://www.mit.edu/~bgrosof/#Consulting

[1] http://www.ruleml.org

[2] http://silk.semwebcentral.org

[3] http://www.mit.edu/~bgrosof/#EBRC2007Talk
Received on Friday, 31 July 2009 22:01:44 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 31 July 2009 22:01:44 GMT