<scribe> scribe: Darko
Paul Vincent is giving presentation on OMG Production Rules
<AlexKozlenkov> Are you aware of a company called Kabira who has a similar MDA driven process with a custom workflow input?
<AlexKozlenkov> http://www.kabira.com/
<sandro> Paul apologized that the slides he's using are not available yet, but he'll mail them out shortly.
<AlexKozlenkov> Question to Paul: can actions be grouped together as a transaction?
<AlexKozlenkov> Paul, any link for the action language? Are communication actions supported?
<sandro> http://www.w3.org/2005/rules/wg/wiki/Rulesystem_Arrangement_Framework
chrisW: RIFRAF is to
characterize
different rule systems
...different types of semantics in
rule systems
Harold: it is a classification system
Gary: it cannot be used for reactive rules at this point
chrisW: we need to extend some
discriminators of RIFRAF to cover other staff we are discussing about
Discussing Syn: Syntactic discriminators
<sandro> (confusion about rifraf 1.1.)
<AlexKozlenkov> no sound
<sandro> sorry -- conversation is moving too fast to move the mic
<AlexKozlenkov> Okay
chrisW: rifraf 1.1 seems to be about scope
<sandro> Harold: Existential variables, anonymous variables
<sandro> Hassan: the head of a rule in LP is sometimes constrained to have all distinct variables
ACTION: Harold to add some examples for rifraf 1.1
<sandro> Harold: forward-chaining woulod generate non-ground-facts, without range restriction
Gary: rifraf 1.2 discriminator for production rules also
<sandro> Axel: typed variables should go here
chrisW: 1.3 is a bit ambiguous
harold: is a classical definition from logic programming
Dave: why is rifraf 3 an interesting discriminator for RIF?
harold: it is interesting to look at the reduction between 2 languages
hasan: this is not what we call conservative extension
sandro: translating to and back would you get the intial rules, or would L-T have been done?
axel: this is a syntactic
discriminator; whether you allow complex formulas or not
michael: these discriminators
have an impact on the rif
csma: rifraf is going to be
used to prioritise requirements.
chrisW: some
of the discrims are not very important for rif. Some are features of
specific languages. rif might not carre about all these features. Later
we will consider which is important for RIF.
Gary: There has to be a more
normal terminology; current terminology does not apply straightfowardly
to PRs.
chrisW: rifraf
5-4 are obvious (syntactic sugar)
michael:
rifraf 6 seems like datatypes
chrisW: it is
purely syntactic
michael: u can
have strings or URIs; I think it is more about datatypes
chrisW: rifraf
6 needs clarification
Discussing SeS: Syntactic-entailing-Semantic
Discriminators
harold: SWRL is an example of a
homegeneous approach. Hybrid approaches AL-log
chrisW: mixing OWL with rules
harold: 2 reffers to rulesets
dave: it
doesn't feel as a discriminator for rule languages but for rule sets
Gary: the
working memory is part of the knowledge
Hasan:if you
want to interchanche PRs you don't interchange the working memory.
michael: why
is 4 interesting?
chrisW: are
there variable-free languages interesting for RIF?
michael: don't
knwo of
harold: gives
example
chris: this is
an example of variable-free rules not languages
Gary: mail
filtering rules
chrisW: this
is questionable
csma: 2.1 and
4.1 are the same
michael: we
have to decide which discriminators are useful
chrisW: let us
understand them first
chrisW: number
a premisses...what is a premisse?
harold: one
premisse is easier to be implemented than two
hassan: are
there languages for discriminator 6?
chrisW:
discriminator 7 whether we can name a rule
harold: label
them and whether u can then use the labels to edit them or reference
them.
...u can use them in a semantic way to assign certain rules priorities
chrisW: we are
going to discuss about the discriminators. some might not be of
interest to RIF
Break