Part of the Workshop Session Notes


Rules Workshop -- Session 4 -- Standards Work

13 May 2005


Sandro Hawke
Eric Miller


<DanC_DCA> (I didn't get that about dinner. did anybody get it?)

OMG PRoduction Rule Representation- an Overview (Paul Vincent)

pv from Fair Issac

pv: members include ilog, ibm, pega, ...

background on OMG, PRR (prod rule representation) status, and how this fits in with rule interop

OMG - most known for UML, CORBA

scribe: vendor, domain org membership
... framework of MDA (model driven arch)
... technologies of UML2
... domains (finance, space, telco, defense)
... rule specific - BEI (BSBR, PRR)
... rule related (BPDM, OMD, OCL)

.me is missing the acronym expansion

OCL == Object Contraint Language

PRR - Metamodel and associated representation

(if <conditions> then <actions>)

organized by rulesets

- sequentially processed rules

- forward chaining inference rules

(most typically used in buisiness rule environments)

Interchance based on XMI (?)

why is the PRR standard required?

- mutliple representations used by industry forthe same concept

industry need to align the business rules approach with UML / OO software dev best practices

co-dev with PRRuleML

.me appreciates the suggestion

State of PRR

- based metamodel defined

(ruleset and rules)

built examples library (URL??)

PRR vs Rule Interop candidates

Proposition: an equiv to PRR, possibly a concrete syntax for PRR, is required for run-time rule exchange

PRR works beyond the web

<timbl> "Beyond the web" ... himm is file: beyond URI ?

Q: is the interpretation the same as OCL?

A: yes

OCL - object constrint language

<DanC_DCA> er... I heard the answer: no, the interpretation is not necessarily the same; we've borrowed the constructs and mutated the semantics as needed

<timbl> "OCL is very bound to objects -- are you limitingto that interp?" -- no.

.me thanks for help
.me cant hear questoipn

Q: does this (PRR?) enable (rational rose, XSB, etc. ) do rule modeling?

A: yes

(arrg.. cant hear)

Q: When is UML2 comming out?

A: almost out now.. the production core is not out and it relates (?) to PRR.. we're expecting this to be out next year

Q: Why arent we considering event condition action rules?

A: general view that ECAR can be a subset of the work we're doing

Q: what does execution mode have to do with rule modelling?

A: it determines the scope of what I can represent with the rules

scribe: also, you can think of this within the model of defining hints to engine

Q: I hear you have to do things "beyond the web"

scribe: could we simply webify the system and more or less leave things unchanged?
... is there something special 'doing things outside the web'?

A: in short, no

scribe: if there is commercial interest, this would be on the checklist


we dont see PRR being special / different for a web standard.. but whatever the interop standard is we'd like to ease this transition from PRR to ...

<scribe> .... (new speaker) ...

Daniel Selman - Java Community and Rule Engine Standards

scribe: hope to bridge gap between business and KR communities
... quite a lot of java rule engines and interest

JSR94 - only java standard in this area

JSR-94 - Java Rule Engine API

-> http://www.jcp.org/en/jsr/detail?id=94

40+ Open Source and commercial java rule engines - http://www.javarules.org

integration with Spring framework

Specification does not define the behavior of the rule engine (explicitly ruled out of scope)

scribe: but remains a contentious issue

equiv JDBC without SQL

waiting for standard underlying rule language

"I want to write a ruleset.. i want to run the same set on jrules and blaze"

Jess is the reference implementation for JSR-94

<DanC_DCA> AT&T Labs Research - R++

Q: why not simply extent the Java languge to support rules (e.g. ATT's R++ to C++)

A: Sun is reluctant to extend java

(missed other answer)

<DanC_DCA> other answer: R++ is not the architecture that Java rules vendors have used

Q: whats the relationship to APIs with what W3C should standardize?


<DanC_DCA> R++ patent

bingo, thanks..

A: in most vendor systems, rule sets are monolythic things.. we can imagine this being more dynamic and recombined from many parts of the web

Q: can you take JSR94 and take 2 different rule sets on 2 diff platforms

A: (thinks the answer is) yes, but at the API level only

(missed question)

<scribe> ... new speaker ...

Donald Chapin, Business Rules Team - Semantics of Business Vocabulary and Business Rules

important to separate business modeling and information system modeling

Donald Chapin, Business Rules Team - Semantics of Business Vocabulary and Business Rules

SBVR: A synthesis of 4 disciplines

- vocabulary standard (ISO 1087 terminology standard)

- business proactices

- formal logics

- linguists and communication

<sandro> SBVR == Semantic Beaver

SBVR - includes among other things logic (FOPL with some limited extensions)

Key SBVR New Contribution - Semantic Formulation

- language for talking about meaning and concepts of rules

a way of structuring the meaning of definitions, rules, question / queries

optimized for peoople and natual language - not for machine processing

<sandro> Deontic and Alethic logics, plus temporal

in the submission document, there are examples in structured english

SBVR - what can it contribute to / require from rule language for interoperability

q: has anyone done the SBVR -> RDF/OWL transformation

a: not yet

q: (more of a statement) if we can have these SBVR -> tranformations in place, this gives us useful set of examples, requirements, input to the Rules community

a: absolutely

<DanC_DCA> "you're establishing a common ground so that your computers can communicate on your behalf"

<DanC_DCA> that ability to say "oh, I don't understand that part, so I'll ignore it" is a core semantic web principle: partial understanding. It seems to clash pretty badly with negation-as-failure, as far as I can tell.

<DanC_DCA> meta-comment: lots of people have looked at this metro diagram on the web and haven't had the benefit of the explanation

<DanC_DCA> timbl: there's a video version for the royal [something]

<DanC_DCA> WWW Past & Future - Berners-Lee - Royal Society

<DanC_DCA> webcast

<DanC_DCA> dan demo'd sparql, starting at the WG page http://www.w3.org/2001/sw/DataAccess/

<DanC_DCA> then going to http://esw.w3.org/topic/DawgShows

<DanC_DCA> then to sparqler http://sparql.org/query.html

<DanC_DCA> c = [ is math:sum of (a b)]

<DanC_DCA> { ?C math:sum (1 2 4) } => { ?C a Answer}

<sandro_post_hoc> scribe: Eric Miller

<sandro_post_hoc> scribeNick: em

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.122 (CVS log)
$Date: 2005/05/13 19:35:55 $