See also: IRC log
sandro: When will we have the (Philippe Bonnard's)
... we don't want to push it off to the end of the meeting.
RESOLUTION: accept minutes of April 17 meeting
APPROVED: accept minutes of April 17 meeting
RESOLUTION: accept minutes of April 24 meeting
APPROVED: accept minutes of April 24 meeting
<StellaMitchell> chris: any liaison reports? ... (none)
<josb> I did not check yet
<csma> I did
<josb> I filled it in myself :-)
<josb> anyway, people will have some time
chris: Any new information on the F2F?
Jos: The room has been booked; the technical needs are taken care of (mikes, phones, etc.).
josb: People should book hotels as soon as possible, because the choice is already starting to get limited.
Chris: why the 3 hotels listed on the wiki page?
josb: Took one hotel from each of
three price categories: each is reasonably near the bus
... However, any hotel on the list would be reasonable.
Chris: How many people have answered the questionnaire?
Chris: 15 people have
... please fill out the form, whether or not you're attending. Need to plan for the meals, etc.
Chris: Philippe will be giving a demo.
<PhilippeB> Conference ID 5792048
Chris: We'll try using the demo using the raindance conferencing software.
Chris: If you've already installed raindance, all you need to do is log in with the conference no Philippe provided above.
<Deborah_Nichols> I am - Deborah
yes, I am
Philippe: We'll have a demo of
POC showing the RIF core enabling interchange between JBoss and
... in order to test the core of RIF and its extensibility
<josb> don't wait for me
Philippe first presented slides showing the POC architecture.
Philippe, will you be posting the slides somewhere on the web, so we can reference them?
<ChrisW> gary, we are in a demo
Philippe: Table containing translations from production rules concepts to RIF concepts
<GaryHallmark> is there a raindance conf id?
<DaveReynolds> I've just been kicked off the Raindance conf
<sandro> Can we see the web page directly, or do we have to only see it like this, through the screen-broadcaster?
<sandro> (If I have a password to MISMO's site, I don't know about it.)
<Allen> what is "<constaints>?
<csma> Sandro, are you on the Mismo wiki?
<sandro> No, I don't know the password to the MISMO wiki.
<csma> you do not need a password: you set it yourself
Philippe: We give examples of ILOG production rules, followed by translations of the set of rules into RIF, realized as an xml document, followed by the translation of the xml document into JBoss. We have a table showing how specific production rules are mapped into the RIF syntax. This table also shows which features are not supported in RIF Core.
<csma> if you are not registered, you register yourself by giving an loggin ID and a password
<ChrisW> daver, did you get back on?
<ChrisW> the conf shows you in
<DaveReynolds> Yes, I had to kill the client and restart
<sandro> Okay, finally got on to the mismo site
<sandro> (free site registration required)
<DaveReynolds> (Requires acceptance of MISMO License terms IPR rules on behalf of employer)
<Deborah_Nichols> I have it just now, too
<csma> Caveat: this is not supposed to be a public wiki. Let us use it for convenience during the demo, but we should as Philippe to copy the relevant info on the RIF wiki...
Philippe: In summary, we showed ideas for
round-tripping between Ilog and JBoss.
... We showed that we went from Ilog to the meta-model and then back to Ilog.
... in more comprehensive roundtripping, we can go from Ilog to xml and back to Ilog; eventually we plan to be able to translate from Ilog to JBoss and back.
Philippe: We can map xml expressions to metamodel and vice versa.
<PaulVincent> Congrats to Philippe and the ILOG / JBoss team for an excellent practical RIF demo!!!
Philippe: By the end of May, at the next meeting, the aim is to have a complete back-and-forth translation of the JBoss and ILog rules engines.
<csma> For some reasons you do not seem to hear me :-(
<csma> but apparently I cannot speak
<csma> Philippe, why the getters?
<Harold> Philippe, what would be the (update) semantics of the setters?
<PhilippeB> Harold, In fact, we change the state of the object and we n otify it to the rule engine.
Issue 25: RDF and RIF Core
Chris: There has been some discussion in email of RDF in Core, most of which has centered around b-nodes.
Sandro: Jos proposed 3 options:
Sandro: 1rst option: Tell people who want to use b-nodes to map to the closest thing they can find in RIF. Essentially that's skolemizing.
<csma> Harold, could you re-ask by email your question about what are the constraints in Philippe's aprocrif, lest we forget to have that discussion?
Sandro: This is the option favored by Sandro and Jos.
<csma> I will ask mine about getters...
Jos: 2nd option: allowing facts with existential variables.
Jos: 3rd option: modularized approach, like in DL+log
Jos: But it's not clear for production
rules how this would be done
... and the semantic implications for extensions, e.g., with negation, are not clear.
Jos: Negation over rule predicates would be different than over DL predicates.
Jos: negation as failure vs.
... 2nd option has problem with extensions that can't account for existential operators, including extensions with negation and extensions with production rules.
Chris: in the first option, b-nodes are replaced with skolem constants.
<csma> Would taht mean no bnodes in Core?
Chris: in the second option, allowing facts with existentially quantified variables, skolemization happens ..
Chris: Is it possible to have a quantifier that scopes the b-node, so that skolemization would replace it with a function containing quantified variables contained therein?
Jos: That depends if you allow rules or facts. If you just allow facts: that situation wouldn't happen.
Sandro: For the third option, skolem constants are sufficient.
Sandro and Jos: discussion about when skolem constants would be sufficient, when skolem functions are needed.
The answer depends on what happens as rules are fired
Chris: Jos is saying that the
idea of going with skolem constants to replace b-nodes only
works when rules languages support published b-node
... is this an example of a rules language that doesn't support the b-node semantics?
Sandro: I don't see the problem.
Chris: Sandro is saying that rules language that supports generating new symbol for each rule firing would know to put in a function rather than a constant.
Jos: But what if a rule fires twice
-with new variable binding?
... vs. rules firing twice, but keeping the same variable binding?
Sandro: This is very much a production rule issue.
Jos: Nothing will prevent people
from exchanging RDF graphs; RDF already has a semantics for
... leans toward option 1, skolemization during translation.
<csma> In the current core, an ANDCONDITION can be an existential
Sandro: same as Jos
Dave: RIF is supposed to be able
to use RDF as data model.
... point of being able to do all these things, including b-nodes, is to be able to translate the triples.
... Need RDF in both head and body of rule.
Jos: But if we embed RDF as a set of
facts, can make usual RIF rules that quantify over facts, or
data in facts. So RDF data gets used by default.
... even though syntactically, we might want to allow people to write things that look like RDF triples in head and body.
<sandro> -1 mirroring SPARQL
<sandro> -1 mirroring SPARQL's treatment of b-nodes
Harold: SPARQL has dealt with b-nodes --- can we use similar treatment to Sparql's treament?
Dave: not following Jos's suggestion.
Chris: I think we're all in
... RDF graph translated into RIF rules language
... RDF rules get translated into RIF core.
(missing above, how does RDF graph get translated into RIF rules language .. missing end of chris's sentence.)
Chris: Michael not here.
Harold: not sure if Michael agrees or disagrees.
Sandro: Jos should propose text
to go into the document regarding this issue.
... Jos should just put the text into the wiki?
<DaveReynolds> +1 to putting in email
Chris: No, Jos should put the text into email.
<sandro> ACTION: JosDeBruijn to propose text in e-mail for changes to RIF Core, encompassing the "just Skolemize" approach to handling b-nodes [recorded in http://www.w3.org/2007/05/01-rif-minutes.html#action01]
<rifbot> Sorry, couldn't find user - JosDeBruijn
<sandro> ACTION: Jos_De_B to propose text in e-mail for changes to RIF Core, encompassing the "just Skolemize" approach to handling b-nodes [recorded in http://www.w3.org/2007/05/01-rif-minutes.html#action02]
<rifbot> Sorry, couldn't find user - Jos_De_B
<sandro> ACTION: Jos_de to propose text in e-mail for changes to RIF Core, encompassing the "just Skolemize" approach to handling b-nodes [recorded in http://www.w3.org/2007/05/01-rif-minutes.html#action03]
<rifbot> Sorry, couldn't find user - Jos_de
<sandro> ACTION: Jos de B to propose text in e-mail for changes to RIF Core, encompassing the "just Skolemize" approach to handling b-nodes [recorded in http://www.w3.org/2007/05/01-rif-minutes.html#action04]
<rifbot> Sorry, amibiguous username (more than one match) - Jos
<rifbot> Try using a different identifier, such as family name or username (eg. jdebruij, jderoo)
<sandro> ACTION: jdebruij to propose text in e-mail for changes to RIF Core, encompassing the "just Skolemize" approach to handling b-nodes [recorded in http://www.w3.org/2007/05/01-rif-minutes.html#action05]
<rifbot> Created ACTION-285 - to propose text in e-mail for changes to RIF Core, encompassing the \"just Skolemize\" approach to handling b-nodes [on Jos de Bruijn - due 2007-05-08].
Chris: Dave had sent email on the topic, April 18
Dave: In my email, I tried to characterize 3 main options and give examples/test case, showing difference between these three.
Dave: I used examples where P and Q
were both individuals and predicates; to see how the 3 options
(disjoint sorts, overlapping sorts, punning (ONDS)) worked out.
... DS --> DS no work
... ONDS --> ONDS, DS, OS, need new names to distinguish between P as an individual and P as a predicate, but otherwise the translation should work okay.
... same with ONDS to ONDS
... the hard case comes when there are overlapping sorts translating to disjoint sorts, to ONDS, or to other overlapping sorts.
<josb> for core I prefer DS
<sandro> Dave: Argument for DS --- it makes translating from RIF into any mode easy.
<josb> OS is not compatible with OWL DL!
Chris: But despite the ease, no one seemed to support the disjoint sorts view.
Sandro: We don't understand the issues well enough to pick sides yet.
<DaveReynolds> josb: sorry, yes I meant OWL/full of course
Chris: So it's still on the table.
Dave: DS could be called no aliasing
Chris: or, no overloading
Dave: Here's an analogy in programming
languages: it like the difference between static and dynamic typing.
... ONDS is like static overloading in Java.
... vs OS --- dynamic overloading, e.g, as in Python.
<sandro> ds: no aliases, no overloading (Early C)
<sandro> onds: punning overloading -- disambiguate at compile time (Java)
<sandro> os: like dynamic typing (Python)
<josb> I understand
<DaveReynolds> I think I do
<sandro> I think I understand them again, now.
<PaulVincent> Understand: "sort of" ...
<Allen> not sure
<AxelPolleres> need more context, but was late again, sory. :-(
Chris: Allen, what would help you understand this better?
Chris: We'll discuss this again next week; everyone should try to read the wiki page and prepare for a discussion next week.
Sandro: We need to look at all
the possible combinations of languages in the mapping.
... (e.g., translating from ONDS to ONDS, OS to ONDS), etc.
<AxelPolleres> same here, will progress starting from next week more
Nothing new for RIFRAF
<PhilippeB> Leora, there is a public Wiki page presenting the global architecture of the MISMO POC: http://wiki.mismo.org/MISMOWiki/Wiki.jsp?page=ProofOfConcept
This is scribe.perl Revision: 1.128 of Date: 2007/02/23 21:38:13 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/XXX/JBoss/ Succeeded: s/YYY/ILog/ Succeeded: s/pUnning/ONDS/ Found Scribe: Leora Morgenstern Found ScribeNick: LeoraMorgenstern Default Present: Philippe_Bonnard, Harold, +39.047.101.aabb, Dave_Reynolds, josb, ChrisW, Deborah_Nichols, PaulVincent, Leora_Morgenstern, StellaMitchell, Sandro, Allen_Ginsberg, csma, johnhall, Gary_Hallmark Present: Philippe_Bonnard Harold +39.047.101.aabb Dave_Reynolds josb ChrisW Deborah_Nichols PaulVincent Leora_Morgenstern StellaMitchell Sandro Allen_Ginsberg csma johnhall Gary_Hallmark Regrets: FrançoisBry IgorMozetic DavidHirtle MichaelSintek MohamedZergaoui HassanAitKaci Agenda: http://lists.w3.org/Archives/Public/public-rif-wg/2007Apr/0100.html Got date from IRC log name: 1 May 2007 Guessing minutes URL: http://www.w3.org/2007/05/01-rif-minutes.html People with action items: b de jdebruij jos jos_de jos_de_b josdebruijn WARNING: Input appears to use implicit continuation lines. You may need the "-implicitContinuations" option.[End of scribe.perl diagnostic output]