Admin

next meeting will be: 12th December

<Deborah_Nichols> I have Christian's issue summary, and I will post it.

<ChrisW> action 171 complete

<ChrisW> action 165 complete

<Deborah_Nichols> If you want to give me an action to post the roundtripping issue, go ahead.

proposed accept minutes of telecon nov14th

minutes accepted

<ChrisW> Resolved: accept Nov 14 minutes

proposed accept minutes of telecon nov21st

<ChrisW> RESOLVED: accept Nov 21 minutes

minutes of nov21 approved

proposed accept minutes of face2face of nov 4,5th

<ChrisW> RESOLVED: accept f2f4 minutes

minutes of face2face of nov 4,5th approved

Technical Design

<ChrisW> action-142 continued (indefinitely)

<sandro> we're looking at http://www.w3.org/2005/rules/wg/track/products/9

<ChrisW> action-159 continued

<PaulaP> continued

<ChrisW> action-160 continued

<ChrisW> action 152 continued

<ChrisW> action 156 continued

<ChrisW> action 157 continued

<ChrisW> action 170 completed

<ChrisW> action 182 continued

<ChrisW> action 184 continued

<ChrisW> action 186 continued

<ChrisW> action 188 continued

discussing web-ized version of core proposal in

<csma> http://www.w3.org/2005/rules/wg/wiki/A.1.0_Nucleus%3A_Positive_Conditions

michaelkifer: document is a modification of previous version

The syntax and semantic is more flexible,

Key concept is removng distinction between constants, predicates and function symbols

And to use sorts to reintroduce distinctions as needed

This allows the use of URIs for constants, predicates etc.,

gives good connection with RDF,

and also allows primitive data types to be introduced

csma: why nucleus positive condition does not have slotted syntax

michael: two issues

slotted syntax can appears as formulae or terms

there is also potentially an issue with the model theory

it appears to be incompatible with conventional semantics

<Harold> Christian, the current lack of slotted syntax in the Nucleus is also a Wiki page maintenance issue.

discussion needed with hassan

<Harold> I expanded the earlier page with slotted syntax, since the Nucleus page was not yet discussed.

<Zakim> sandro, you wanted to ask if this is talking about using URIs for individual integers or for the type/class of all integers

<Zakim> sandro2, you wanted to ask for clarification on difference between terms "core" and "nucleus"

sandro: do you mean an URI for each integer?

michael:uri for sorts

sandro: what is the diference between core and nucleus

michael: just to distinguish two documents' proposals

<sandro> MichaelKifer: I'm just using the term "nucleus" here to distinguish the document from the "core" document -- the choice of term is not semantic

alex: are we interested in round tripping to f-logic?

alex: not nec. web-ized - core

michael: nucleus means not yet a core

<sandro> MichaelKifer: ALSO the core is going to be bigger than this, so maybe the nucleus is smaller than the core.

slotted syntax for terms/formula

csma:We will discuss slots later please

daveR: could you explain why sorted approach is tied into web-ization

michael: it is not directly, it just helps
... if you introduce a wall, as is normal between predicates and constants it is hard to remove later

without the wall, you can re-introduce something similar with sorts

platform for extensions

e.g. we can decide later to allow dates as predicates or not

csma: you seem to make URIs semantic not syntactic feature

michael: URI is syntax not semantic

hassan: michael is proposing a fundamental idea of using sorts to partition the space of names
... term-level and preciate-level constructs are treated similarly

chrisw: trick not yet fully dealt with

leave to the semantics issues of what a URI really is

michael we can use multi-sorted logic to add additional layers to encode our desires

daveR: I am concerned about the entanglement of semantics and syntax

globally unique names is separate from types

<Hassan> Multi-sorted logic allows to separate signatures (i.e., sets of symbols) making up your expressions. The semantics of a sort is simply that of a domain of value.

michael: a URI is a string in a particular format

<Hassan> Sorts may have internal structures

there will be a URI sort, with a syntax and obvious semantics

other sorts may not be permitted to be predicates

chrisw:These sorts are not necessarily user defined

<Hassan> Sorts do not denote necessarily *finite* sets!

csma: not clear that sorts are syntactic only

chrisw: semantics not yet clarified

<Hassan> ChrisW is correct!

michael: in multi-sorted logic, sorts are part of the syntax

conventional semantics in terms of herbrand base

sorts are subsets of the universe

uris are simply syntax

URIs are analogous to identifiers

csma: we decided that global identifiers must be URIs

chrisw: URIs are tokens in the syntax

michael: we can have rules to define where things can be used

chrisw we can use sorts to clarify the role of particular identifiers

csma: the rules for URIs seem dificult to write

michael: we have to specific the syntax for each sort

csma: if an identifier is of a type X then it must have syntax Y

<sandro> MichaelKifer: Yes, the Herband Universe includes relations. This is a generalization of FO logic. By introducing the notion of sorts, you allow yourself to define FO logic.

FrankMcCabe: If you're putting relations into the domain of discourse, this is not First Order logic.

sorted logic is syntacally not first-order but the translation in first-order logic is trivial.

MichaelKifer: You just define a "sort" of predicates, boolean, which gives you FO logic.

<Harold> Frank, you can split the universe into just the the common sorts of 1st order logic: Con, Fun, and Rel.

FrankMcCabe: we don't WANT to generalize FO Logic. We don't need to do that.

<Francois> sorted logic is ***NOT*** higher-order.

<AlexKozlenkov> It is not HO logic

<Francois> Higher-order required that ones quantifies over all possibnle sets - what Michael proposes is not of this kind.

<Francois> +1 with Hassan's explanations.

<sandro> Hassan: as long as you never synthesize relations, it's not Higher Order.

<Harold> Even if we allow Hilog/RDF-like querying of Rels, it's still only a conservative extension of 1st order logic.

<sandro> MichaelKifer: The semantics remains first order because it can be mapped into First Order Logic.

<Frank McCabe> ACTION: francis to clarify problem with sorts [recorded in http://www.w3.org/2006/12/05-rif-minutes.html#action03]

<rifbot> Created ACTION-191 - Clarify problem with sorts [on Francis McCabe - due 2006-12-12].

<Harold> Semantically, even Hilog is still 1st order logic.

csma: is this the right way to web-ize the syntax?

is it enough?

chrisw: it allows us to web-ize the syntax, it still needs to be done

we need to define the rules

michael: we need to define the rules

csma: is this at the abstract syntax level, or concrete syntax level?

michael: probably at both levels

alexK: need to remind you that we handle java to rif also

rule set may be completely defined wrt local namespaces

csma: no word yet on mapping

alexK: mappings needed

csma: we will continue with approach for now

michael: concerns should be noted in email

<rifbot> Created ACTION-192 - Update CORE document with multi-sorted appraoch [on Harold Boley - due 2006-12-12].

<DaveReynolds> What "current document"?

<ChrisW> current doc: http://www.w3.org/2005/rules/wg/wiki/CORE

csma: rif conditional language in core document
... slotted syntax

what is the meaning on the slots

hassan: I was trying to make it more tutorial

can re-express regular unification in terms of constraints

e.g. CLP gives you logic with added benefits

also put in a lot of references to handle objects with constraints

Is next week Ok?

csma: a little late

<rifbot> Created ACTION-193 - Circulate document explaining CLP [on Hassan Ait-Kaci - due 2006-12-12].

hassan:I will do my best

slotted notation is a notation for constraints

confusion between domains and data models

can use constraints to represent domains, terms, graphs etc.

any data model can be represented using constraints

slotted notation is a notation for constraints

csma: if you use constraints to represent a data model can it be re-interpreted as pattern matching and contraint solving

hassan: constraints allows you to capture the semantics of data models etc.

constraints represents a unifying formalism

csma: will using constraint mech to isolate the data model allow rules intoerchange between e.g. Prolog and e.g. object model language like ilog

hassan: yes

csma: would like an example

alexK: rountripping between f-logic and herbrand?
... roundtripping between production rules and prolog also?

gerd wegner's approach allows roundtripping

hassan: constraint also allows it

csma: will continue discussion of extensibility by email

F2F

csma: we have no alternate proposals for f2f 5

f2f 5 will be at Mitre in last week of feb

<sandro> PLEASE ANSWER THIS SURVEY: http://www.w3.org/2002/09/wbs/38457/f2f5-schedule/

csma: will you attend if it is two or three days

Liason

csma: Paul Vincent will be new liaison for PRR
... Thankyou to Axel for study on sparql and rif

<PaulVincent> PRR liaison: no news

UC&R

<ChrisW> action 144 continued

<ChrisW> action 72 continued

allenG: whether or not we include concept of translation should be equivalent

csma: equivalence is implied already

allenG: why not may it clearer
... you can say something is equivalent without writing hte software

csma: action is continued

allenG: cannot see an exit strategy at the moment

chrisw: how many proposals are there

what are the proposals?

paulV: let us review again next week

action 167 continued

action 168 continued

action 169 continued

action 181 continued

csma: what is the status of the UCR document?

allenG: sent out an email to editors to revise sections according to f2f
... issue wrt what phase I, II requirements mean
... on track to have UCR done early next year

csma: Do we need to map to specific actions for specific people?

<PaulVincent> Qu: where are the use cases numbered?

csma: people already agreed at the f2f

actions on people who did requirements on use cases

csma: they are listed in the minutes

<DaveReynolds> What about open issues for UCR?

RIF-RAF

action 148 continued

<PaulVincent> Continued

action 149 continued

action 172 continued

<LeoraMorgenstern> continued

action 173 continued

action 174 continued

action 175 continued

action 176 continued

action 177 continued

<PaulaP> continued

action 179 continued

action 187 continued

<Csma> need to make more progress on actions

<Hassan> perhaps should we be less prompt to bestow actions without realistic deadlines?

hassan: there should be priorities attached to action items

deadline should be negotiated more realistically

hassan: action deadlines should be carefully decided

people involved in technical design and rifraf and ucr should be prioritized

AOB

<Harold> How long will the seasonal break be? Dec 19 last telecon 2006? Jan 9th first telecon 2007?