Task points to analyse condition language of commercial rule engines and contact JSR-94 people

Guys

Here are some details about my tasks from the last teleconference.

1. I have contacted Daniel Selman through one channel but he has not yet
responded. For those of you who do not know him, he used to be with ILOG
then moved over to BEA and is know for having been the Specification
Lead of JSR-94 and the founder of the website javarules.org. He also
co-created the rules engine built into BEA Portal product. The idea is
to get the announcement about the current RIF Draft to be communicated
to the JSR-94 community in order to ensure that those of them who are
not yet participating in RIF, become more involved at least by providing
their feedback.

2. I am looking into four languages with a goal to find common points
between their condition languages and the condition language outlined by
Harold Boley et al. The languages I have chosen--hopefully not
contradicting someone else's earlier commitment--are JBoss Rules, SUN
Glassfish Self-Management, Mozilla XUL, and Prova (I'm the author of the
latter so I will be able to say a few things more). The Jboss Rules
(former Drools) is a fairly classic production rules system with syntax
and semantics that so far look to me as mappable to Frame logic. Now we
haven't used Frame logic directly in our documents. However, we should
be aware that Frame logic is essentially a syntactic sugar on top of
logic programming with negation based on well-formed semantics. I don't
see Jboss Rules using WFS but a scoped NAF seems to be there as well as
the use of quantifiers and connectives on the RHS. XUL is interesting in
that it actually integrates RDF querying. Sun Glassfish Self-Management
is possibly the most limited and not so well described.

3. Independently from that, I'm surveying the existing languages with a
view to extend the condition language WIKI with some analysis of
possible action types in active rules. There is a surprising variety of
action literals (effectors) and I must stress that these effectors may
not necessarily by the head of a rule. A more general ECA style would
put actions into the special action compartment while the approach used
in BPEL and Prova would allow actions (like communication actions) to be
present more or less freely.

More later,

Alex Kozlenkov

Advanced Technologies Group
Betfair Ltd.
www.betfair.com
W3C AC-Rep

Received on Friday, 14 July 2006 09:40:26 UTC