% notes from JC telecon 6/1/04
% by Benjamin Grosof

participants:

Sandro Hawke
Harold Boley
Benjamin Grosof
Norman Sadeh
Pat Hayes
Said Tabet
Mike Dean

moderator:  Benjamin Grosof, since Mike Dean had to join late due to traveling

%%%%

agenda:

Followup from DAML PI Meeting and SWSI F2F in NYC late May:
especially on services requirements for rules,
especially on developing a FOL superset of SWRL/RuleML.  

A. Benjamin, Mike, Norman, and Harold who had been there, 
gave a summary outbrief:

[Slides about rules and services from those meetings, by Benjamin and Mike D.,
soon will be (or already are) available, stay tuned to the mailing list
for more about that.]

SWSL requirements on near-term (summer/fall) include:
1. LP with nonmonotonic negation/priorities, for policies and contracts and 
        data mapping (i.e., the "SCAMP" group of tasks)
    - preferably as superset of SWRL (where OWL/DL ontologies are 
        restricted to DLP)
    - later also preferably with extensions towards HiLog and 
            probably F-Logic-y syntax
2. general FOL for complex procedural process modeling
    - preferably as superset of SWRL

Both of these go beyond current SWRL.
(1.) is mainly met by the current full RuleML
(2.) is not met by current full RuleML, but was requested by DAML to 
       be addressed by the Joint Committee, in a "80-20" mode within 2 months,
         not waiting upon SCL effort but coordinating with it and DRS

RuleML has all along had a plan to extend its syntax to FOL, 
but has been waiting on the SCL effort.  

A presentation syntax (i.e., human-editing string syntax roughly
similar to Prolog or N3) is also needed for progress in the DAML Rules Tools
effort.

%%%%

Discussion about Action Plan to develop a FOL superset of SWRL

There was discussion, not recorded blow by blow, but here are the 
highlights of the consensus.  Pat contributed much background about
current status of SCL including its expressive features and syntactic
approach.

Let's proceed to define a syntax for classical FOL as a 
superset (extension perhaps with some modifications) 
of SWRL's Horn rules part, i.e., of SWRL's RuleML part.

Once that is defined, we would like to extend it further
(perhaps not within the 2 months, however) to the more general case
tackled by SCL, in which predicates can also be in the range of quantification,
in manner roughly of HiLog or OWL-Full, which can be translated 
[says Pat] into classical FOL.

Also desirable, timing-wise probably before that further expressive extension,
would be to use something like the approach from SCL where one can point
at other syntactic encodings of FOL formulas/knowledge, indicating that
they are other encodings or have been translated from other encodings.  

In connection with that, there was discussion about to what degree is it
problematic to have multiple ways to express the same formula within 
the same syntactic language, e.g., the one we are developing.

We don't have a name yet for all this, the working name for the moment
is "SWRL-FOL"
- Another alternative:  SWFOL (pronounced "Swiffle") [Benj]

We should define the syntax in the following sequence of steps:
1. abstract syntax ("abstract syntax" here in the manner/sense 
       of current SWRL and earlier OWL doc's)
2. XML syntax/info-model
3. RDF syntax/info-model

wrt representing unorderedness of children, perhaps using Relax-NG might help
the specification overcome the awkwardness of XML-Schema in this regard
[Sandro, others]

Action Plan further details:

Pat will provide a pointer to latest SCL information 

Strawman proposals will be created and presented, starting next telecon,
hopefully distributed by email by Monday 6/7

Volunteers to create strawman proposals:

Pat -- has some ideas based on SCL especially

Benjamin, Said, Harold, Norman -- in part based on RuleML