RE: [RIF] [UCR]: What is the RIF (revisited) --> changing vendor rule languages

From: Michael Kifer [mailto:kifer@cs.sunysb.edu] 
Sent: Thursday, February 09, 2006 6:23 PM
To: Vincent, Paul D
Cc: public-rif-wg@w3.org
Subject: Re: [RIF] [UCR]: What is the RIF (revisited) --> changing
vendor rule languages 

 

 

 

"Vincent, Paul D" <PaulVincent@fairisaac.com> wrote:

> 

> Michael Kifer Wrote:

> 

> > "Vincent, Paul D" <PaulVincent@fairisaac.com> wrote:

> > >

> > > There are certainly similarities between the rules market today
and the

> > > SQL market of the early 80s. This is why the vendors are
supporting the

> > > OMG PRR and RIF efforts. However, so far there is zero push for

> > > additional rule construct support in the rule languages. 

> > 

> > Perhaps those customers don't know better?

>  

> Possibly, but more likely their needs are simple enough to be
addressed

> by the main commercial vendors.

>  

> > But ask companies like Ontoprise, OntologyWorks, XSB Inc., etc., to
get a

> > different view.

>  

> Well, these are a different class of rules platform (ontology-based

> rules). A perfectly valid class, but one that seems to have a much

> smaller market share today.

 

In 1982, SQL market was infinitesimally small compared to other DB

products.  And those competitors were arguing that the needs of their

customers are perfectly met by hierarchical DB products. They also used
to

claim that relational DBs will never fly because they are
unimplementable,

non-scalable, cause leprosy and impotence. Are we hearing the same
arguments

again? ;-)

 

If I recall, the commercial relational DB offerings were endorsed by the
academic community. I don't remember the academic world trying to
compete with Oracle and Ingres. :-) 

 

> > > > Perhaps the reason why the rules

> > > > market is fairly small is because the current commercial rule
languages 

> > > > are so pathetically poor and ill-founded.

> > > 

> > > If this is the case I have not seen any evidence to support it,
and would

> > > welcome any links to support this hypothesis!

> > 

> > I dunno about the links. But I can tell you about my own limited
experience

> > consulting for companies. (Consulting is not what I do regularly or

> > eagerly, hence the disclaimer.) Twice my clients needed a rule
language and

> > twice we considered various commercial products, including some from

> > companies represented in RIF. And twice we decided to use something
else,

> > non-commercial and open source. The commercial offerings just didn't
cut it.

> 

> I can quite believe it if the need was to reason over an ontology.

> 

 

No, not ontology reasoning. A fairly simple reasoning, which could be
done

easily with simple Prolog, but was hard to control with forward-chaining

production rules.

 

Yes: another example would be a goal-driven problem that is best handled
using backward chaining, that might be awkward in a forward chaining
environment. 

 

> I will offer you a beer at the next F2F to discuss further!

 

Wine. I'll take that beer next time, when the F2F will be in Belgium :-)

 

You're on!

 

 

      --michael