[RAF] [Ontology] Report of task force conference call, with subsequent email

On Wednesday, the RAF ontology task force held a conference call to set
the agenda and propose a preliminary division of labor. I forwarded a
summary of this conference call to the members of this task force, and a
discussion about various points of the conference call ensued. We are
moving this discussion to the RIF mailing list, since that is where it
belongs. Attached please find the initial summary, along with the
discussion so far.

Best regards,
Leora


---------- Forwarded message ----------
Date: Wed, 11 Oct 2006 17:17:40 -0700 (PDT)
From: Leora Morgenstern <leora@steam.Stanford.EDU>
To: Allen Ginsberg <aginsberg@imc.mitre.org>, Hassan Ait-Kaci
<hak@ilog.com>,
     Sandro Hawke <sandro@w3.org>, Frank McCabe <frankmccabe@mac.com>,
     Axel Polleres <axel.polleres@urjc.es>
Subject: summary of today's meeting



I am planning to email the text below to the RIF mailing list. Let  me
know if this looks okay.

Leora
-----------------------------------------------------------------------------

Following below is a summary of the initial  teleconference for the RIF
ontology task force (11 October 2006):

Participants: Allen Ginsberg, Sandro Hawke, Frank McCabe, Leora
Morgenstern, Axel Polleres

1. Defining the objective of the task force:
============================
There was some discussion on defining the objective of the task force.
The consensus seemed to be that we are to make explicit the ontology
that is implicit  in --- in Frank's terms, "baked into" ---  the current
RIFRAF document (
http://www.w3.org/2005/rules/wg/wiki/Rulesystem_Arrangement_Framework )
and the questionnaire (
http://www.w3.org/2002/09/wbs/38457/RAFQuestionnaire )  .

(It was  also noted  that there is not at this point a 1-1
correspondence between the RIFRAF document and the questionnaire, and some
corrections/modifications have to be made to both in order to align them with
one another.)

Frank further suggested that what we are doing can be seen as defining
an ontology of rules languages.

2. First step for the task force
======================
We decided as a first step, in developing this ontology, to divide up
the classes of discriminators discussed in  the RIFRAF document. More
precisely:
(a) Each member of the task force will take one section of the RIFRAF
document and construct an ontology of rules languages based on this
section
(b) The output of each member will be an ontology in OWL. We may use a
tool like Protege to produce this ontology.
(c) Our task, in developing this sub-ontology, will be not only to map
what is explicit in the RIFRAF document but also to try to make
explicit what is implicit, to clarify questions, and to provide commentary.

It is likely that this exercise will bring to light new problems and
issues, such as the conditional dependencies between the classes.

3. Next steps for the task force
=======================
Once we do this exercise, we will have to merge these subontologies.

Also, we will have to relate this work to previous work on ontologies
of rules languages, including the exising work in ML. Harold Boley, as
one of the authors of that work, will be consulted in this regard.


4. Division of labor
=============
The current division of labor is as follows:

Hassan: Section 2  ---- Syntactic-entailing-Semantic Discriminators
Allen:     Section 3  ---- Semantic Discriminators
Sandro:  Section 4  ---- Pragmatic Discriminators
Leora:    Section 5  ---- Discriminators for Event-Condition-Action (ECA) Rules
Frank:    Section 6  ---- Types in rule languages

Section 1 will be done after the other sections have been completed.

Axel will review the ontologies. He is most suited for this role since
he was the original author of the questionnaire.
-----------------------------------------------------------------------------------------------------------------------------------
---------- Forwarded message ----------
Date: Thu, 12 Oct 2006 15:30:11 +0200
From: Hassan Aït-Kaci <hak@ilog.com>
To: Leora Morgenstern <leora@steam.Stanford.EDU>
Cc: Allen Ginsberg <aginsberg@imc.mitre.org>, Sandro Hawke
<sandro@w3.org>,
     Frank McCabe <frankmccabe@mac.com>, Axel Polleres
<axel.polleres@urjc.es>
Subject: Re: summary of today's meeting

Leora Morgenstern wrote:

> I am planning to email the text below to the RIF mailing list. Let
me
> know if this looks okay.
>
> Leora
> [...]
> We decided as a first step, in developing this ontology, to divide up
the classes of discriminators discussed in  the RIFRAF document. More
precisely:
> (a) Each member of the task force will take one section of the RIFRAF
document and construct an ontology of rules languages based on this
section
> (b) The output of each member will be an ontology in OWL. We may use
a tool like Protege to produce this ontology.
> (c) Our task, in developing this sub-ontology, will be not only to
map what is explicit in the RIFRAF document but also to try to make
explicit what is implicit, to clarify questions, and to provide commentary.
> [...]
> The current division of labor is as follows:
>
> Hassan: Section 2  ---- Syntactic-entailing-Semantic Discriminators

Thanks Leora for the minutes. I think the approach sketched is a sound
one.
I also would be glad to participate as you are suggesting, however with
the caveat emptor that (1) I do not understand what the phrase means
and
(2) I only know OWL from what I have read in the few academic languages
describing it and its foundations. I have no experience using software
based on it. I could get and learn Protege but I have little time, in
addition to my being on the road right now and until the F2F4.

However, the nature of the ontologies that arise from the
questionnaires
seems to need but a concept hierarchy (tree or lattice) with a few
functional
attributes denoting deicriminating values. I can start sorting through
my
sub-part of the questionnaire (i.e., syntactic descriminators).

BTW, Frank is right: we *are* trying to derive a rule classification
ontology. This is why I wonder why there are no RuleML aficionados
(e.g. Harold Boley, et. al.) joining in this task force so that we
may avoid reinventing the square wheel. Moreover, what goes under
Syntactic-entailing Discriminators what originally proposed by him
as inflenced by RuleML. Perhaps it would be godd to involve him in
the task force as well (at least on this part).

Finally, what time frame are we talking about here (to come up with
the sub-ontologies)?

In conclusion, I think that this is a good plan with the reservations
that I have expressed regarding the topic assigned to me, my fluent
"hooting" skills  (i.e., speaking OWL) and lack of time to learn until
the end of this month, and finally the need to involve RuleML folks for
the part they are mostly responsible for.

Cheers,

-hak
-- 
Hassan Aït-Kaci
ILOG, Inc. - Product Division R&D
tel/fax: +1 (604) 930-5603 - email: hak @ ilog . com
---------------------------------------------------------------------------------------------------------------------------------
---------- Forwarded message ----------
Date: Thu, 12 Oct 2006 10:59:31 -0400
From: "Ginsberg, Allen" <AGINSBERG@imc.mitre.org>
To: Leora Morgenstern <leora@steam.Stanford.EDU>,
     Hassan Ait-Kaci <hak@ilog.com>, Sandro Hawke <sandro@w3.org>,
     Frank McCabe <frankmccabe@mac.com>, Axel Polleres
<axel.polleres@urjc.es>
Subject: RE: summary of today's meeting

Hi Leora,

Nice write up.

I do have a concern (point of clarification) regarding the notion that
we are producing an "ontology of rule languages."  That could be taken
to imply that we are going to classify all the existing instances of
rule languages. Perhaps ultimately we will, but I didn't think that was
the initial task at hand.  I would describe our current task as using
an ontology language to specify (possibly define) and inter-relate the
system of concepts surrounding the idea of a rule language.  The
starting point is the set of discriminators given in the RIFRAF (and
the questionnaire), but in order to elaborate those concepts it might
be necessary to add others.

In any event, are we agreed that at this point we are not going to
create or use instances, e.g., JRules, SWRL, etc. in our formal
ontology?

Allen



--------------------------------------------------------------------------------------------------------------
---------- Forwarded message ----------
Date: Thu, 12 Oct 2006 11:28:04 -0400
From: Sandro Hawke <sandro@w3.org>
To: "Ginsberg, Allen" <AGINSBERG@imc.mitre.org>
Cc: Leora Morgenstern <leora@steam.Stanford.EDU>,
     Hassan Ait-Kaci <hak@ilog.com>, Frank McCabe
<frankmccabe@mac.com>,
     Axel Polleres <axel.polleres@urjc.es>
Subject: Re: summary of today's meeting


We should probably be having this discussion on the WG mailing list,
not
just among us.   But, um, I guess I'll reply anyway.   We can re-play
this for the WG I guess.

> Nice write up.
>
> I do have a concern (point of clarification) regarding the notion
that
> we are producing an "ontology of rule languages."  That could be
taken
> to imply that we are going to classify all the existing instances of
> rule languages. Perhaps ultimately we will, but I didn't think that
was
> the initial task at hand.  I would describe our current task as using
> an ontology language to specify (possibly define) and inter-relate
the
> system of concepts surrounding the idea of a rule language.  The
> starting point is the set of discriminators given in the RIFRAF (and
> the questionnaire), but in order to elaborate those concepts it might
> be necessary to add others. =20

That sounds about right, although I was say "rule system" not rule
language.  There's a chicken/egg duality there, so maybe there is no
point in making a distinction, but I think when push comes to shove
we'll find that the imagined characteristics of languages are often
different from the real characteristics of the systems that supposedly
implement those languages.  ("The difference between theory and
practice
is greater in practice than in theory.")  RIF needs to provide
interchange between real, deployed systems -- not theoretical
languages.

Maybe, in truth, the ontology needs to cover both.

     ISO_Prolog_Language_Version_X is a language with certain
characteristics.
     SWI_Prolog_System_Version_Y is a system which mostly implements
                           ISO_Prolog_Version_X and has these
characteristics.


> In any event, are we agreed that at this point we are not going to
> create or use instances, e.g., JRules, SWRL, etc. in our formal
> ontology?

Answering the questionaire is equivalent to creating instance data like
this, right?

   -- Sandro
--------------------------------------------------------------------------------------------------------
---------- Forwarded message ----------
Date: Thu, 12 Oct 2006 12:06:55 -0400
From: "Ginsberg, Allen" <AGINSBERG@imc.mitre.org>
To: Sandro Hawke <sandro@w3.org>
Cc: Leora Morgenstern <leora@steam.Stanford.EDU>,
     Hassan Ait-Kaci <hak@ilog.com>, Frank McCabe
<frankmccabe@mac.com>,
     Axel Polleres <axel.polleres@urjc.es>
Subject: RE: summary of today's meeting

Hi Sandro,

> In any event, are we agreed that at this point we are not going to
> create or use instances, e.g., JRules, SWRL, etc. in our formal
> ontology?

    >> Answering the questionaire is equivalent to creating
instance data like
    >> this, right?

    >>   -- Sandro


I suppose it is.  Do we have access to the questionaire data at this
point?

Allen
---------------------------------------------------------------------------------------------------------
---------- Forwarded message ----------
Date: Thu, 12 Oct 2006 12:17:47 -0400
From: Sandro Hawke <sandro@w3.org>
To: "Ginsberg, Allen" <AGINSBERG@imc.mitre.org>
Cc: Leora Morgenstern <leora@steam.Stanford.EDU>,
     Hassan Ait-Kaci <hak@ilog.com>, Frank McCabe
<frankmccabe@mac.com>,
     Axel Polleres <axel.polleres@urjc.es>
Subject: Re: summary of today's meeting


> I suppose it is.  Do we have access to the questionaire data at this
> point?

Yes.   WG members can see it; it's linked from the questionnaire.

   http://www.w3.org/2002/09/wbs/38457/RAFQuestionnaire/results

I can also do a SQL dump of it, if anyone wants to try to automate the
conversion.

     -- Sandro

---------- Forwarded message ----------
Date: Fri, 13 Oct 2006 07:29:32 +0200
From: Axel Polleres <axel.polleres@urjc.es>
Reply-To: axel@polleres.net
To: "Ginsberg, Allen" <AGINSBERG@imc.mitre.org>
Cc: Leora Morgenstern <leora@steam.Stanford.EDU>,
     Hassan Ait-Kaci <hak@ilog.com>, Sandro Hawke <sandro@w3.org>,
     Frank McCabe <frankmccabe@mac.com>
Subject: Re: summary of today's meeting

Ginsberg, Allen wrote:
> Hi Leora,
>
> Nice write up.
>
> I do have a concern (point of clarification) regarding the notion
that
> we are producing an "ontology of rule languages."  That could be
taken
> to imply that we are going to classify all the existing instances of
> rule languages. Perhaps ultimately we will, but I didn't think that
was
> the initial task at hand.  I would describe our current task as using
> an ontology language to specify (possibly define) and inter-relate
the
> system of concepts surrounding the idea of a rule language.  The
> starting point is the set of discriminators given in the RIFRAF (and
> the questionnaire), but in order to elaborate those concepts it might
> be necessary to add others.
>
> In any event, are we agreed that at this point we are not going to
> create or use instances, e.g., JRules, SWRL, etc. in our formal
> ontology?

I would agree... these instantiations should follow from the filling
attampts in the questionaire... which we should consider at some point!

axel

Received on Friday, 13 October 2006 11:24:13 UTC