W3C RDF Core Working Group IRC Chat Logs for 2002-06-18

04:06:01 Disconnected from irc.w3.org (ERROR :Closing Link: logger_1[~rdfcore-l@tatooine.ilrt.bris.ac.uk] (Ping timeout))

04:07:03 Topic now RDF Core F2F HP Labs, Bristol

04:07:03 Users on #rdfcore: logger_1 em

07:38:01 Disconnected from irc.w3.org (ERROR :Closing Link: logger_1[~rdfcore-l@tatooine.ilrt.bris.ac.uk] (Ping timeout))

08:31:05 Topic now RDF Core F2F HP Labs, Bristol

08:31:05 Users on #rdfcore: logger_1 guha JosD mdean dajobe-la gk bwm em

08:31:08 <dajobe-la> start of meeting

08:31:53 <dajobe-la> ACTION danbri: republish to the existing rdf schema namespace uri the latest rdfs schema

08:32:11 <dajobe-la> objections from PatrickS recorded

08:32:44 <dajobe-la> review of agenda

08:32:50 <dajobe-la> added Graph Syntax

08:52:23 Disconnected from irc.w3.org (Connection reset by peer)

09:41:50 Topic now RDF Core F2F HP Labs, Bristol

09:41:50 Users on #rdfcore: logger_1 bwm_ danbri jjc-scrib guha JosD mdean dajobe-la gk em

09:42:21 <jjc-scrib> path: are we talikng about the value spac or the lexical space

09:44:17 <jjc-scrib> jjc: this is untidy!

09:44:31 <jjc-scrib> jjc: how do we know which facets to use for "10"

09:44:52 <jjc-scrib> sergey: we use the datatype range of the property

09:45:08 <jjc-scrib> path: you seem a little disengeneous

09:45:56 <jjc-scrib> path: we focus on the difference between number and strings

09:46:09 <jjc-scrib> path: on the one hand you say they are the same thing

09:46:58 <jjc-scrib> path: we still have to decide which facets to use

09:47:42 <jjc-scrib> path: connolly entailemtns

09:48:51 <jjc-scrib> gk: how does pat make the connolly entailment fail under sergey?

09:49:03 <jjc-scrib> bwm: who is compelled by sergey's discussion

09:49:07 <jjc-scrib> jos me

09:49:50 <jjc-scrib> jos I liked it because .... (in no more than two sentences)

09:50:00 <jjc-scrib> the semantics of datatypes are captured by facets

09:51:51 <jjc-scrib> conversation between guha and sergey

09:57:16 <jjc-scrib> patrick xsd subsumes your framework

09:57:16 <jjc-scrib> more discussion defeating the scribe

09:57:16 <jjc-scrib> frank: this is interesting

09:57:16 <jjc-scrib> frank: we haven't seen enough details so this proposal is not worthy of consideration

09:57:40 <jjc-scrib> frank: I would like to see more details

09:58:11 <jjc-scrib> guha: bwm is so charming

10:25:37 <jjc-scrib> guha filled in a table

10:26:20 <jjc-scrib> L L+Pred L+Pred+Subj L+pred+DT L+pred+subj+dt

10:26:38 <jjc-scrib> 1: y y y y y

10:32:51 <dajobe-la> more filling in the table

10:33:22 <jjc> jjc is now known as jjc-scrib

10:33:44 <jjc-scrib> jjc was only one wanting df depending on subj

10:34:33 <jjc-scrib> sergey and jos want dependency on L only

10:37:49 <jjc-scrib> down to considering L versus L+pred+DT

10:38:12 <jjc-scrib> 2b (connolly entailment film title "10") differents

10:40:13 <jjc-scrib> 4 cannes entailment L no, L+pred+DT, yes

10:41:19 <jjc-scrib> cc-pp can just about cope with L and L+pred+DT

10:41:29 <jjc-scrib> patrick's stuff wont work with L

10:41:47 <jjc-scrib> path DAML+OIL is not comaptible with L

10:44:56 <jjc-scrib> sergey: cannes implementation requires implementation effort

10:47:10 <jjc-scrib> straw poll: L 3,

10:47:33 <jjc-scrib> L+pref+DT: 5

10:47:51 <jjc-scrib> undecided: 3

10:50:38 <jjc-scrib> path: L keeps implementators lives simpler

10:50:52 <jjc-scrib> path: L+pred+DT end-users lives gets easier

10:53:05 <jjc-scrib> revote

10:53:10 <jjc-scrib> L: 2+ connolly

10:53:26 <jjc-scrib> L+pred+DT: 5

10:54:11 <jjc-scrib> L+pred+DT: 6

10:54:17 <jjc-scrib> concur: 2

10:56:07 <jjc-scrib> sergey: procedurally change to untidy semantic is substantial change

10:56:21 <jjc-scrib> generl: current MT versus older MT

10:57:02 <jjc-scrib> path: the MT has had three drafts each with different opinion

10:59:19 <jjc-scrib> jos: if you wnat a fucntion write a function, use a bnode.

11:02:11 <jjc-scrib> bwm: we should suggest to the community and ask for feedback

11:02:38 <jjc-scrib> bwm: suggest the simplext

11:03:01 <jjc-scrib> jjc: but what do we mean by simplest. Simplest machinery or simplest entailments

11:03:53 <jjc-scrib> frank: community does not know space of possible solns

11:04:13 <jjc-scrib> frank: perhaps you should put the most complicated mechanism and see if the community complain

11:05:47 <jjc-scrib> can we ask the community in a two page write-up with each proposal

11:06:04 <jjc-scrib> guha: aaraon will have something to say

11:06:56 <jjc-scrib> bwm: the question is do we want 2(b) or the cannes entailment

11:12:11 <jjc-scrib> sergey is worried that people will not understand implications of choice

11:12:21 <jjc-scrib> frank: we get to review message before it si sent

11:12:32 <jjc-scrib> frank: sergey can also post on rdf-interest as well

11:13:00 <jjc-scrib> patrick: I am concerned that we repeat the discussion on rdf interest

11:13:48 <jjc-scrib> jjc: should we use voting tech on W3C site

11:13:53 <jjc-scrib> general response: no

11:16:55 <jjc-scrib> bwm: not universal consensus that e-mail to rdf interest good, but some consensus

11:17:58 <jjc-scrib> 8 in favour of quick e-mail, carried

12:11:20 <gk> -------------------------------afternoon session---------------------------

12:11:48 <gk> gk is now known as gk-scribe

12:12:55 <gk-scribe> Mike Dean's talk tomorrow 10:00AM, Guha's at 11:00AM

12:14:01 <gk-scribe> Datatypes -- actions:

12:14:42 <gk-scribe> ACTION bwm, draft email soliciting community input on datatyping

12:15:50 <gk-scribe> -----Graph syntax-----

12:16:46 <gk-scribe> Agreed yesterday: properties don't have namespace, hence namespace not in graph.

12:17:14 <gk-scribe> Question: can isolated nodes exist in graph?

12:17:16 <dajobe-la> jjc's graph doc http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002May/att-0120/01-RDF-XML_Syntax_Specification__Revised_.htm

12:17:23 <dajobe-la> announced in http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002May/0120.html

12:18:20 <gk-scribe> Reasons to not allow: (a) can't express in N-triples, (b) complicates concept of equality

12:19:27 <gk-scribe> PatH noted that people use isolated nodes to label a graph; jjc notes having a dummy type node has similar effect

12:19:52 <gk-scribe> s/node/arc/ above

12:20:05 <gk-scribe> PatH noted that people use isolated nodes to label a graph; jjc notes having a dummy type arc has similar effect

12:22:02 <gk-scribe> DECIDED: no isolated nodes in graph

12:23:12 <gk-scribe> Charater normalization in graph -- what about XML literals? Jeremy proposes to allow any XML literal in well-formed graph, not note that interperability problems may result if not using XML normalization rules

12:24:32 <dajobe-la> graph syntax note "Note: If compatibility with XML version 1.1 is desired, then XML fragments in RDF graphs must be restricted to those that are fully normalized according to [XML 1.1].

12:24:32 <dajobe-la> A"

12:26:25 <gk-scribe> DECIDED: jjc proposed text on character normalization is approved

12:26:46 <gk-scribe> ACTION daveb, fold jjc text into syntax document

12:27:10 <dajobe-la> issue rdfms-assertion

12:27:25 <dajobe-la> gk: after last telecon discussion...

12:27:40 <dajobe-la> social aspects of rdf hasn't gone into the desc of rdf so far

12:27:50 <dajobe-la> ... got some text

12:28:22 <dajobe-la> social dimenson, MT talks of possible worlds, but there is also a human aspect

12:28:44 <dajobe-la> and also social process of publication e.g. putting an rdf file on a home page, you might assume asserted versus test cases which might not be assumed so

12:28:51 <dajobe-la> social context of use

12:29:18 <dajobe-la> truth of rdf statements

12:30:50 <dajobe-la> less formal description of interpreting truth

12:31:43 <dajobe-la> guha: willg et some objections that rdf apps should just do rdf interpretaitons

12:32:03 <dajobe-la> PatH: also will object too, some terminology too

12:32:12 <dajobe-la> ... tweaking maybe

12:33:20 <dajobe-la> gk: trying to hilight text I've got

12:33:23 <dajobe-la> as sw evolves, will get more formally deduced assertions

12:34:08 <dajobe-la> ... what does it mean to assert a graph

12:36:22 <dajobe-la> reading text from gk

12:38:38 <dajobe-la> When an RDF graph is asserted in the web, its publisher is saying something

12:38:38 <dajobe-la> about their view of the world. (The mechanism for deciding whether or not

12:38:38 <dajobe-la> a graph is asserted is not discussed here, but it is presumed that the

12:38:38 <dajobe-la> publisher's intent will be clear in some way -- social convention or

12:38:38 <dajobe-la> logical deduction.) There are a couple of socially grounded ideas

12:38:39 <dajobe-la> here: the publisher of a graph, and the social content in which it is

12:38:43 <dajobe-la> published.

12:38:49 <dajobe-la> ---

12:39:01 <dajobe-la> discussion of above para from gk's text

12:39:05 <dajobe-la> -- http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Jun/0178.html

12:43:38 <dajobe-la> guha: people as well assa writing speech, write in artifical languages such as tables of figures, pictures and mean to assert them; rdf is a lnaguage liek that

12:43:52 <dajobe-la> PatH: some test cases are implicit ...

12:44:26 <dajobe-la> such as implication by using rdfs reasoning on separate statements to give a conclusion

12:45:01 <dajobe-la> ... other test case is when rdfs:comment is used to say "this is the class of all people"

12:45:19 <dajobe-la> it should not entail that rdfs:Human is type; no enailment follows from the comment

12:46:01 <dajobe-la> brian: legal binding - lets stay away

12:46:25 <dajobe-la> re GET in HTTP says it has no side effects, so GET can buy no books from a web bookshop

12:46:45 <dajobe-la> in our context, we have to say what an rdf document means, but that's all

12:47:13 <dajobe-la> PatH: conclusion of statements is an rdf thing, if it is libellous/rude is a legal thing

12:48:31 <dajobe-la> bwm gives example of whether a mail body is asserting it (yes), vs an attachement (not?)

12:49:07 <dajobe-la> PatH: not let people hide behind not asserting valid rdf conclusions

12:49:22 <dajobe-la> bwm: we aren't going to say attachments aren't asserted, that is in the mail spec

12:49:30 <dajobe-la> but it is our business to say what rdf spec means

12:49:48 <dajobe-la> danbri: if we make that distrinction, should make things easier

12:50:34 <dajobe-la> PatH: whatever the wrapper, the entailed valid rdf conclusions must be made

12:50:50 <dajobe-la> danbri: mixed namespaces

12:51:33 <dajobe-la> .. are a rats nest

12:54:16 <dajobe-la> PatH: using a vocab sort of accepting meaning by composor of terms?

12:54:26 <dajobe-la> .. what entailments from what I right am I sort-of commited to?

12:54:42 <dajobe-la> ... if you say you publish in rdf/rdfs, you are buying in to it

12:54:45 <dajobe-la> danbri: but layering

12:55:25 <dajobe-la> FrankM: but extra entailments might come from people who understand vocab

12:56:18 <dajobe-la> jjc: desc of term may be libellous, using the term is just a uri-ref

12:56:40 <dajobe-la> PatH; but using the term and agreeing to it might be libelous?

12:56:45 <dajobe-la> but comment has no semantics in MT

13:05:39 <dajobe-la> discussion of mime types for rdf

13:05:42 <dajobe-la> and assertions

13:11:33 <dajobe-la> discussion of publishing and always asserting them

13:13:45 <dajobe-la> refering to words from cannes meeting

13:16:53 <dajobe-la> maybe need to indicate ourur thinking in some document

13:21:55 <gk-scribe> AGREED: aijm to put a form of the Canes statement in a normative document (TDB), and a copy of GK text, suitably edited, into the primer

13:22:56 <gk-scribe> ACTION gk, to contnue working on primer text

13:23:07 <gk-scribe> ACTION bwm, ensure the Canes text goes into a normative document

13:24:43 <gk-scribe> AGREED: issue closed

13:25:03 <gk-scribe> --------Lbase--------

13:33:21 * em waves, thanks scribe(s) for there notes and wishes he could be there in person

13:33:42 <dajobe-la> hope things are going ok, em

13:34:01 * gk-scribe hi em

13:39:08 <JosD> Guha talking about http://tap.stanford.edu/SemanticWebSemantics.html

13:39:22 <dajobe-la> announced http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Jun/0049.html

13:41:06 <gk-scribe> -----Guha talks about Lbase-----

13:41:15 <gk-scribe> -----Guha talks about Lbase-----

13:42:15 <gk-scribe> Background: current approach is that each SW language has its own model theory. But model theories are diffcult and are not machine-consumable.

13:42:34 <gk-scribe> An alternative approach is axiomatuic semantics.

13:43:22 <gk-scribe> Lbase document defines semantics for a logical language that can underpin a range of SW languages, with mappings for each language

13:43:33 <gk-scribe> Lbase has its own model theory.

13:44:46 <gk-scribe> Proposal is to publish mapping from RDF to Lbase as a non-normative (appendix?) of RDF semantics document

13:45:03 <gk-scribe> keeping the RDF model theory as normative.

13:46:32 <gk-scribe> odel theory vs axiomatics semantics is not a matter of taste: one is not better - they address different goals

13:47:43 <gk-scribe> Technical point: description ogics are not strictly contained in FOL, because they have cardinality constraints.

13:48:46 <gk-scribe> Lbase may be enhanced in the future in ways that do not affect the definition or semantics of RDF

13:50:21 <gk-scribe> There are also proof theoretic semantics and algebraic semantics. We're not going there. (?)

13:50:42 <bwm_> Concern expressed to ensure that the whole community with an interest in the layering issue has an opporuntity to contribute and make their arguments

13:52:12 <gk-scribe> Arguments in favout of Lbase: addresses layering issue (needs wider review); more accessible for many people - on this basis makes sense to include in RDFcore document

13:53:23 <gk-scribe> Another advantage: we can have just one concept of entailment across all Sweb languages

13:55:19 <gk-scribe> jjc: propose that Lbase be in a separate document, because we already have semantics for RDF and that satisfies our charter requirements

13:55:50 <gk-scribe> DanBri: the charter doesn't stop us looking at other issues.

13:56:02 <danbri> from charter, [[

13:56:02 <danbri> The RDF Core group must take into account the various formalizations of RDF that have been proposed since the publication of the RDF Model and Syntax Recommendation. The group is encouraged to make use both of formal techniques and implementation-led test cases throughout their work.

13:56:03 <danbri> ]]

13:56:25 <danbri> danbri: (not that I want to obsess on charter, but just as data point)

13:56:43 <bwm_> s/not/note/

13:56:47 <bwm_> :)

13:57:13 <danbri> (either works :)

13:58:42 <gk-scribe> FrankM: notes that Lbase facilitating layering doesn't require others to use it. Argues that Lbase may be useful, but doesn't need to be normative.

14:00:18 <gk-scribe> Guha: (in response to Q from Jos) Lbase is machine readable, but not intended to be used for publishing on the semantic web

14:01:26 <gk-scribe> Guha: Reason for not using KIF: it's much more non-standard than Lbase.

14:02:55 <gk-scribe> Sergey: notes that Lbase is useful for exploring semantic issues (e.g. bnodes, datatyping).

14:04:42 <gk-scribe> Sergey: we need to understand how Lbase positions w.r.t. other documents

14:05:17 <gk-scribe> PatH: presence of (RDF-using-)Lbase shouldn't materially affect other RDFcore documents

14:06:23 <gk-scribe> jjc: no major rewrite possible at this stage -- precluded by "process issue" (?)

14:07:58 <danbri> re reference to DAML-S formlization work and limits of current DAML MT: see http://lists.w3.org/Archives/Public/www-ws/2002Jun/0002.html

14:08:11 <gk-scribe> Discussion of whether to put out model theory. Guha says yes - certain modelntheories have well understod algorithmic practice: having a model theory is good for some purposes.

14:10:30 <gk-scribe> Decisions... (assuming no technical show-stoppers)

14:11:58 <gk-scribe> DECISION: description of RDF using base will becaome a non-normative appendix of RDF formal semantics document (was RDF model theory) if it is ready in time for publication. The description of Lbase will be published as a W3C note.

14:14:31 <gk-scribe> ACTION DanBri, will work with Guha/PatH to get Lbase document published as W3C note

14:14:52 <bwm_> morning aaron

14:15:00 <AaronSw> g'morning

14:16:33 <dajobe-la> hi; the logs might not be all up to date due to network problems at UoB

14:17:08 <gk-scribe> ACTION PatH, rename RDF model theory to RDF formal semantcis, and include RDF Lbase semantics as non-normative appendix. (jjc opposes this)

14:18:40 <gk-scribe> DaveB proposes move graph syntax into formal semantics document

14:20:54 <gk-scribe> jjc proposes that a general RDF "odds&sods" document be provided

14:21:07 <gk-scribe> Guha suggests generic vocabulary document

14:22:30 <gk-scribe> DanBri: latest "schema" document talks about vocabularies, to avoid some loaded terms

14:25:51 <gk-scribe> GK re-raises idea of "umbrella" specification.

14:28:50 <gk-scribe> Proposed "Resource Description Framework" document, with graph syntax, what it means to assert, character normalization, other material from primer...

14:29:59 <gk-scribe> Choice: expand current schema, or have new document?

14:34:58 <gk-scribe> DECIDED: will write new document "Resource Description Framework"

14:35:13 <gk-scribe> ACTION GK, prepare outline for document

14:36:03 <gk-scribe> Editors will be GK and JJC

14:36:20 <gk-scribe> -- documentation rules

14:36:54 <gk-scribe> jjc proposes "series editor" with responsibility for consistency across documents: proposes Brian

14:37:52 <gk-scribe> DECIDED: will be series editor

14:38:34 <gk-scribe> ACTION BWM, collate responses concerning style and propose guidelines for document style

14:49:04 <dajobe-la> (conformance)

14:49:12 <gk-scribe> ACTION BWM, consider whether conformance statement is needed

14:50:01 <gk-scribe> Discussion of implementation hints? (e.g. is full implementaton of canonicalization needed?) Suggested that FAQ would be appropriate.

14:52:35 <gk-scribe> DaveB: conformance section might help exit from CR

14:53:04 <gk-scribe> ACTION BWM, start discussion of CR exit criteria

14:53:32 <AaronSw> (wow, i go away for a minute and we have a whole new spec)

14:53:45 <gk-scribe> (Passing test cases may be a part of exit criteria, but not sufficient)

14:55:01 <gk-scribe> Updating schema at current locations...

14:56:12 <gk-scribe> Docrules issue: need to check nature of references -- use generic "latest version", or "dated version"?

14:56:30 <em> gk-scribe, et.al... is this new doc assumed to be normative?

14:56:48 * gk-scribe em, yes (I think)

14:57:06 <em> gk-scribe, please ask the group and confirm

14:59:05 <gk-scribe> ACTION BWM,Consider what the current-version link of model and syntax should point to

14:59:28 <gk-scribe> ---break---

15:06:33 <AaronSw> hm? 11:06:04 <jjc-scrib> guha: aaraon will have something to say

15:06:55 <dajobe-la> :)

15:07:07 <dajobe-la> you were there in spirit

15:07:11 <AaronSw> heh

15:07:22 <dajobe-la> but it was datatypes

15:07:45 <dajobe-la> and we decided to ask rdf-ig, logic for feedback on two approaches

15:07:50 <dajobe-la> +others of course

15:07:59 <AaronSw> cool

15:09:35 <dajobe-la> we seem to have closed all issues, mebbe not datatypes

15:09:42 <dajobe-la> (my comment, not official)

15:16:05 <gk-scribe> ---documetn structure done---

15:16:34 <gk-scribe> --------------schedule-------------

15:16:52 <gk-scribe> Primer 26-Aug

15:17:05 <gk-scribe> Syntax 26-Aug

15:18:10 <AaronSw> (what are these dates for? CR? last call?)

15:19:34 <gk-scribe> RDF document: outline by 21-Jun, 1st internal draft by 28-Jul

15:19:35 <dajobe-la> proposed for last call

15:20:53 <gk-scribe> Reviewing likelihood of document ready for last call 26-Aug

15:21:01 <gk-scribe> Primer: OK

15:21:14 <gk-scribe> Syntax: OK

15:21:25 <gk-scribe> RDF "framwork":" tight

15:21:41 <gk-scribe> Formal semantics OK

15:21:49 <gk-scribe> Test cases OK

15:21:52 <gk-scribe> Schema OK

15:23:06 <gk-scribe> Lbase: OK

15:30:27 <gk-scribe> Datatypes: the key dependency is a definitive answer on tidiness by 1st August; if that's done, formal semantics can go, and the datatypes can follow in the time it needs.

15:45:59 <gk-scribe> Discussion of literal form (string,language,XML) stuff... is giving PatH some concern

15:49:19 <gk-scribe> Original issue was that lang is not in the graph: but M&S makes it clear that it's there. TimBL (original issue raiser) doesn't like resolution but accets it if it is truly a clarification of what M&S actually says.

15:49:47 <gk-scribe> Do we really have a reason to reconsider this (literal-structure) issue at this time?

15:53:22 <gk-scribe> jjc: a reason to change this is that the revised handling of datatyping provides a bnode in the graph to hang the language tag. The logicians present don't like this.

15:56:37 <gk-scribe> Procedural proposal: ask Guha/Pat to consider atomic/non-atomic literals at the same time as issues ralating to tidy/non-tidy literals.

15:57:45 <gk-scribe> Brian: the attempt to reopen the issue has faled. PatH has concerns about "new" information. We need to let PatH consider this.

15:58:24 <gk-scribe> ACTION PatH, to evaluate the currently proposed literal form and determine whether it can be easily accommodated.

15:58:38 <gk-scribe> ---------meeting ends----------