% notes on Joint Committee telecon 9/14/2004
% by Benjamin Grosof

participants:

%%%%

agenda:  

REWERSE PPSWR [1] report (Benjamin - 10 min)

FOL RuleML Updated [2] (Harold - 30 min)

SWRL FOL packaging and publication options (all - 10 min)

[1] http://www.pms.ifi.lmu.de/PPSWR04/

[2] http://www.daml.org/listarchive/joint-committee/1792.html

%%%%
PPSWR04 report by Benj:

a REWERSE event
(Benj is on the board, and was an invited speaker as was Michael Kifer)

decent quality
lots of interest in rules / logic programs and ontologies;
also other topics such as temporal reasoning 

REWERSE plans:
next PPSWR04, in Sept. 2005, will be at Dagstuhl, not colocated with ICLP
there will be a summer school, on SW rules etc., in Malta late July 2005

Mike D.:  also relevant:
- DAML Time ontology has existed for a while:
- SWAD-Europe efforts

%%%%

FOL RuleML design discussion: 

Harold walked thru his/the latest draft of FOL RuleML ([2] above)

label -- at top level

forall "closure" within an implies, at top level 
- can be inherited from parent expression 

assert -- at top level
query -- at top level

implies; could have also impliedby and omit head, body tags



Discussion:  


o detailed syntax spec 
 
Benj:  want monolith BNF and XSD/DTD in the document 

Peter: yes, we really need the syntax spec'd out in the document 

(more discussion about this point being an urgent priority) 

Harold:  yes, I volunteer to do that within the next few days

Benj:  want also answerset -- at top level -- e.g., bindings not just
facts  

Harold:  yes


o query and answerset and conclusions 

Benj:  there's an issue of existentially-bound answers cf. DAML Query
Lang and OWL-QL.  For simplicity, let's omit that in first version, then
add as an extension later.  

Harold:  sounds OK.  


Benj:  wrt assert:  what about distinguishing conclusions vs. premises? 
That's an important distinction for LP, esp. with NAF, and we may want to
have it for congruence in the FOL syntax as well.  This is related
to answerset as well, since the answerset to a query is essentially
of the conclusions flavor.  This is related to the Turnstile element
that has been proposed in RuleML V0.8xx.  

Harold:  (missed recording his answer, he discussed Turnstile in part.)  

Peter and Harold:  let's omit turnstile on purpose here, so that we don't 
talk about truth in the language.  
- this was discussed on last week's telecon  

Benj:  in query, I think it would be convenient to name a premise rulebase
that is associated with a query, or that is associated with an answerset,
or to name the query that is associated with an answerset. 
All that could wait til later, after we get the assertional
sublanguage.  However, it would help implementers to have it, as we have
been discovering in our current SweetRules effort on architecting
interfaces for RuleML inferencing tools.  

Mike Dean:  W3 is working on a query language, we should look at what
their design approach is

(more discussion of issues; Ian and Peter and Mike felt that we should
wait on query etc. stuff, Benj and Said ) 

consensus:  since all the query/conclusion/answerset stuff adds complexity, 
and is not what we originally were tasked with by DAML PI Mtg request,
let's avoid doing query etc. for now, and just focus on the assertional 
language.  

o SWRL builtins and other expressive distinctions not already in RuleML 

Benj:  wrt SWRL aspects:
want union of expressiveness, e.g., the data valued properties, builtins

Said:  we can handle the semantics of the builtins via fixed extensions,
similar to in SWRL V0.6.  

Benj and all:  yes


o Negation element in FOL RuleML 

Benj:  let's use Neg (Neg cf. in RuleML V0.85) rather than NOT as in the
current draft 
- rationale:  reuse Neg (from other RuleML sublang's and design principle), 
and cf. standard notion of classical negation in the KR literature

(more discussion of rationale) 

consensus: yes, let's use Neg