W3C RDF Core Working Group IRC Chat Logs for 2002-06-17

06:01:06 Users on #rdfcore: @logger_1

08:43:09 <JosD> Brian elcomes all of us

08:44:19 <dajobe-la> dajobe-la has changed the topic to: RDF Core F2F HP Labs, Bristol

08:44:47 <JosD> Roll Call: FrankM, GrahamK, SergeyM, DaveB, DanB, Guha, PatH, JosD, MikeD, JeremyC, Brian

08:45:57 <dajobe-la> last wg meeting irc logs http://ilrt.org/discovery/chatlogs/rdfcore/2002-06-14.html

08:46:04 <JosD> === 1. list of open issues

08:49:59 <JosD> 8: outstanding issues - 8 left

08:51:18 <JosD> rdfs-xml-schema-datatypes rdfms-contexts rdfms-seq-representation rdfms-assertion rdfs-editorial rdf-namespace-change faq-html-compliance rdfs-isDefinedBy-semantics

08:52:14 <JosD> DaveB proposes to start with the easiest one: rdfms-seq-representation

08:54:51 <JosD> ACTION on Danbri to update the RDFS spec w.r.t. this resolution

08:58:29 <JosD> the d

08:58:49 <danb_lap> Notes: we are putting this stuff in the rdf namespace; we should take care to warn/flag/etc this in the Status of Doc section(s).

08:59:33 <JosD> Path: explain carefully that the usual constraints on lists are not imposed by the semantics

09:00:31 <JosD> RESOLVED rdfms-seq-representation

09:01:30 <JosD> DaveB propose the next one faq-html-compliance

09:03:40 <JosD> Issue Summary: The RDF FAQ suggests how RDF meta data might be included in HTML. The suggested approach is fails HTML 4.01 and XHTML validation

09:05:52 * danb_lap notes that http://validator.w3.org/check?uri=http://xmlns.com/foaf/0.1/ has no validator errors (an xhtml doc with rdf inside it)

09:06:19 <JosD> Jeremy: the page (containing RDF) will will not validate

09:08:51 <dajobe-la> my proposal http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Jun/0099.html

09:11:16 <JosD> ACTION on DanB to update the FAQ to reflect this resolution

09:11:20 <JosD> ACTION on DanB to update the FAQ to reflect this resolution

09:11:47 <JosD> ACTION on DaveB to update the syntax spec to reflect this resolution

09:12:15 <JosD> issue is now closed

09:12:48 <JosD> RESOLVED faq-html-compliance

09:13:20 <JosD> next one: rdfs-isDefinedBy-semantics

09:19:08 <JosD> Guha: it's a subproperty of seeAlso and says, hey, look there

09:19:49 <JosD> proposed text from PatH in http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Jun/0147.html

09:24:16 <danb_lap> newsflash aside: http://news.bbc.co.uk/sport3/worldcup2002/hi/matches_wallchart/mexico_v_usa/newsid_2045000/2045340.stm [[

09:24:17 <danb_lap> Brian McBride, who spent large periods of the match watching from the half-way line as Mexico pushed forward, gave the USA the lead after just eight minutes

09:24:21 <danb_lap> .

09:24:21 <danb_lap> ]]

09:26:16 <JosD> FrankM: how do you draw the boundaries around what is the schema?

09:27:22 <JosD> GrahamK knocks down a text which is now proposed

09:31:37 <JosD> Brian propses a testcase

09:33:57 * danb_lap tries an example: <rdfs:Property rdf:id="myAuthor"><rdfs:subPropertyOf><rdfs:Property rdf:about="http://purl.org/dc/elements/1.1/creator"><rdfs:isDefinedBy rdf:resource="http://purl.org/dc/elements/1.1/"/></rdfs:Property></rdfs:subPropertyOf></rdfs:Property>

09:34:15 <JosD> PatH: (even before Brian finished) NO, it doesn't entail anything but itself

09:35:16 <JosD> Guha says that 'usedAccordingTo' would better express the idea

09:38:37 <JosD> ... oops, forgot that JanG came in an hour ago or so...

09:44:30 <JosD> the original intent was that is didn't assert the triples in the schema

09:51:11 <JosD> Bran summarizes the options: 1/ drop it 2/ add import 3/ keep weak 4/ keep strong

09:52:27 <JosD> cross 4/ and 1/

09:53:10 <JosD> DECIDED 3/

09:54:04 <JosD> ACTION on PatH to update the MT in that isDefinedBy is a subPropertyOf seeAlso

09:54:10 <dajobe-la> This property indicates a resource which contains information about

09:54:10 <dajobe-la> the subject. Often, this property is used to indicate the source of

09:54:10 <dajobe-la> the subject uriref, where its owner specifies its intended meaning.

09:54:10 <dajobe-la> The subject node of this property can be any uriref, and the value

09:54:10 <dajobe-la> may be any document or resource; the usage is not restricted to a

09:54:12 <dajobe-la> particular form or schema

09:54:32 <dajobe-la> -- from http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Jun/0147.html

09:54:41 <danb_lap> 'this does not have any logic meaning'

09:56:35 <dajobe-la> original rdfs said "The most common anticipated usage is to identify the RDF schema given a name for one of the properties or classes defined by that schema. "

09:58:08 <JosD> ACTION on DanB to update the chama spec reflecting this decision and PatH text

09:58:35 <JosD> CLOSED rdfs-isDefinedBy-semantics

09:58:52 <JosD> --------------- coffee break

10:32:35 <danb_lap> ---regroup'

10:33:09 <danb_lap> discussion of 'In RDF, each predicate used in a statement must be identified with exactly one namespace, or schema' in M+S

10:34:12 * danb_lap wonders what sense of 'or' is intended

10:40:51 <danb_lap> action: danbri to add to RDFS 'status of this document' comment w.r.t. 'exactly one namespace or schema' and isDefinedBy.

10:41:30 <JosD> namespaces aren't in the model

10:48:00 <JosD> JeremyC proposes to postpone new issue..., but Brian says that namespaces are not in formal model

10:52:09 <JosD> ... therfore we are dropping the text "In RDF, each predicate used in a statement must be identified with exactly one namespace, or schema." rom M&S

10:52:30 <bwm> therefore there is no special representation of the relationship between a property and its namespace in the rdf model.

10:55:30 <JosD> issue rdf-namespace-change

10:55:37 * dajobe-la points at URIs for W3C namespaces doc http://www.w3.org/1999/10/nsuri

10:56:16 <JosD> RESOLUTION is not to modify the namespace

10:56:18 <danb_lap> what I heard: discussion of 'changing the namespace'; we're adding new stuff into the existing M+S namespace 1999 URI, instead of migrating to a new URI for that namespace. Status of docs will reflect this, and call for implementor feedback.

10:56:28 <JosD> RESOLUTION is not to modify the namespace

10:56:42 * danb_lap nods

10:56:58 <danb_lap> (where modify = change to new uri)

10:57:18 <JosD> CLOSED rdf-namespace-change

10:58:10 <JosD> issue rdfs-editorial

10:58:52 <danb_lap> PROPOSED: RDFS editorial task list does not deserve a separate open issue list.

10:59:07 <danb_lap> RESOLVED. Issue now closed.

10:59:28 <danb_lap> ACTION: danbri to work outstanding editorial todos into next WD.

10:59:55 <danb_lap> ...reflecting individual issues / msgs listed in rdf issue list against RDFS.

11:00:12 <JosD> ACTION on DanB to update the schema spec to reflect rdfs-editorial

11:01:26 <JosD> CLOSED rdfs-isDefinedBy-semantics

11:02:12 <JosD> ==== dark triples

11:03:30 <JosD> ...oops 2 lines before that should have been CLOSED rdfs-editorial

11:04:20 <JosD> Guha goes to the white board and starts his exposition

11:13:40 * danb_lap notes that ssh tunnel to w3 may be volatile (ie. my other one just collapsed)

11:14:58 <JosD> PatH: entail the thing encoded, not the encoding

11:17:06 <JosD> Guha's points: 1/ triple vs assertion 2/ monotonic

11:17:47 <JosD> ... i.e. 1/ triple =/= assertion

11:18:46 <JosD> ... 2/ monotonicity inside layers, accross layers, in triples & assertions

11:21:17 <JosD> ... A |= B then adding anything to A should not let go that away

11:24:38 * PatrickS comes in

11:26:59 <danb_lap> guha: 'sticking w/ MT approach, core of problem.. if you describe the syntax of the logic in the logic itself, you get into well known problems, paradoxes...'

11:27:41 <danb_lap> ...'the constraint of encoding owl and cwm etc using rdf is leading to problems'

11:28:09 <danb_lap> jeremy: 'we're tellign the rest of the world to represent everything in triples, except ourselves?'

11:28:23 <danb_lap> pat: ...you can't syntactically use it for everything, and semantically use it for everything

11:34:07 <JosD> Guha refers to http://www.openhealth.org/WOWG/DTTF

11:39:23 <JosD> ..very hot discussion hard to summerize...

11:42:27 <JosD> PatrickS: reserved predicate is trigger for asserting, why is that monotonic?

11:43:10 <dajobe-la> discussion of how to designate that a triple is not an assertion

11:43:15 <dajobe-la> such as by marking a triple

11:45:35 <JosD> Brian: where are we?

11:46:38 <danb_lap> (trying to capture gist...) "Question is whether we need a mechanism to indicate those triples that are merely encoding syntax for logical expressions, and not encodings of propositions about the world."

11:46:39 <danb_lap> ??

11:47:08 <gk> ... and mechanism must satisfy Guha's 3 monotonicities

11:47:34 <JosD> Jeremy explains why he is opposing dark triples: 1/ no compelling paradoxes 2/ no proven robustness

11:50:46 <JosD> ... 2/ w.r.t. the magnitude of the change

11:56:46 * gk hi eric

11:57:24 <JosD> DaveB: problem versus solution

11:57:51 <dajobe-la> problem (from jon borden msg), requirements for solution and then solution

12:00:12 <JosD> Guha's 1/ + OWL's RDF syntax => turn of things in rdf semantics

12:01:59 <JosD> .. i.e. an argument to have a mechanism to turn of RDF semantics

12:05:41 <danb_lap> --

12:05:52 <danb_lap> part 2 of guha's discussion...

12:05:58 <JosD> RESOLUTION we will try to provide a mechanism such that certain triples in an RDF graph are not having assertional impact

12:06:09 <danb_lap> "monotonicity inside the layer."

12:07:40 <JosD> RESOLUTION 2a monotonicity inside layers

12:08:54 <JosD> ... should always be the case

12:09:36 <gk> 2a: if A |=rdf B then A+x |=rdf B

12:10:19 <JosD> RESOLUTION 2b monotonicity across layers should be possible

12:10:56 <gk> 2b: if A |=rdf B then A |=swel B (must be possible)

12:11:19 <JosD> ----------------------------------- lunch break---------------------------------------

13:05:09 <dajobe-la> resuming after lunch

13:05:43 <dajobe-la> continuing dark triples...

13:05:49 <dajobe-la> dajobe-la is now known as daveb-scr

13:05:54 <daveb-scr> daveb-scr is now known as db-scribe

13:06:38 <db-scribe> guha gives an overview of proposals to do DTs

13:06:47 <db-scribe> proposal 1 - reserved vocab

13:07:05 <db-scribe> and a function isRDFreserved(URI-ref)

13:07:17 <db-scribe> and w3 arch requirement that no network access is required

13:07:24 <db-scribe> suggestion for new uri schema in order to do this

13:07:53 <db-scribe> rather than rely on a list of terms and a registry (online)

13:08:14 <db-scribe> major weakness - new uri schema

13:08:19 <db-scribe> proposal 2 - syntax addition

13:08:29 <db-scribe> every occurrance of a DT be syntanctically marked

13:08:44 <db-scribe> in every syntax - xml, graph, ntriples, ...

13:08:53 <db-scribe> pro - doesn't involve uri schema

13:09:10 <db-scribe> cons - every user has to remember to do this marking on DTs

13:09:28 <db-scribe> versus only designers need to know this and can but it in schemas

13:09:34 <db-scribe> --

13:11:12 <db-scribe> jjc asks should we/can we make our list constructs dark?

13:11:35 <db-scribe> does it make snse to do log:daml:first dark? is it separate from daml:first ?

13:11:42 <db-scribe> (askes PatH)0

13:12:06 <db-scribe> guha - different thing

13:13:59 <db-scribe> discussion of darkening daml list, since that is what an example of what webont wanted DTs for

13:19:29 <db-scribe> jjc shows an example of a daml list and triples that are dark

13:19:40 <db-scribe> i.e the triples with first, rest properties

13:23:40 <db-scribe> brian gives an example of darkening a triple later on...

13:24:43 <db-scribe> existance of uris

13:28:56 <db-scribe> brian summarizes

13:29:36 <db-scribe> can we choose one of these solns?

13:30:44 <db-scribe> syntactic mechanisms

13:30:57 <db-scribe> PatH - triple colouring after jonb

13:31:11 <db-scribe> but that is more than just darkening

13:31:16 <db-scribe> and layering

13:31:55 <db-scribe> JosD: third soln when triples on different levels

13:32:22 <db-scribe> by dereferencing to get different sets of triples

13:32:33 <db-scribe> mechanism could be that they are in a different place (URI)

13:34:13 <db-scribe> dark graphs

13:34:24 <db-scribe> new mime type?

13:34:42 <db-scribe> gk: n3 formula?

13:34:51 <db-scribe> josd: not necessary

13:35:30 <db-scribe> guha: mixing theory and metatheory

13:35:44 <db-scribe> PatH: really extenions to rdf syntax

13:35:59 <db-scribe> JosD: not necessarily via multiple layers that are not mixed (separate grpahs)

13:40:34 <db-scribe> I suggested using a schema to specify darkness was least worst way

13:40:43 <db-scribe> Guha: but someone coujld change it later

13:42:10 <db-scribe> gk suggested using a single uri#frag and registering frag

13:49:54 <db-scribe> PatrickS asks us to consider syntax soln 2

13:52:43 <db-scribe> FrankM asks about darkening existing well known predicates

13:52:52 <db-scribe> ... and that would work with #2

13:55:10 <db-scribe> miked; froma use rpoint of view, 1 and 2 are similar, have to decide ach time what to write

13:55:43 <db-scribe> miked: own and dowl - "night owl"

14:01:06 <db-scribe> example of a property log:http://ww.../owl/foo

14:01:16 <db-scribe> could be used as a subject/object of rdf statements

14:01:19 <db-scribe> but not as a property

14:02:31 <db-scribe> so can be used in a triple <log:http://.../owl/foo> rdfs:Label "The foo..."

14:06:15 <db-scribe> PatrickS proposes webont use xml literals with rdf encoding

14:08:15 <db-scribe> sergey discusses dark documents/graphs

14:13:23 <db-scribe> brian summarises

14:15:22 <db-scribe> proposals

14:15:25 <db-scribe> 1. reserved vocab

14:15:42 <db-scribe> function isRDFdark(uri-ref)

14:15:58 <db-scribe> 2 syntax addition; marking triple(s) as dark

14:16:14 <db-scribe> 3 parsetype literal to allow layers above to store XML

14:16:30 <db-scribe> 4 dark documents, dereferencing

14:16:58 <db-scribe> 5 nested subgraphics (similar to 4, with syntax changes)

14:18:12 <db-scribe> PatrickS outlines an rdf-in-rdf idea

14:20:05 <db-scribe> and Q on higher layers, hiding stuff - they have to put in their syntax

14:20:13 <db-scribe> a darkening mechanism

14:21:31 <db-scribe> decided not 3

14:33:15 <danb_lap> Can we use this:

14:33:16 <danb_lap> [[

14:33:17 <danb_lap> When an RDF processor encounters an XML element or attribute name that is declared to be from a namespace whose name begins with the string "http://www.w3.org/TR/REC-rdf-syntax" and the processor does not recognize the semantics of that name then the processor is required to skip (i.e., generate no tuples for) the entire XML element, including its content, whose name is unrecognized or that has an attribute whose name is unrecognized.

14:33:17 <danb_lap> ]]

14:33:19 <db-scribe> straw polls on various things

14:33:19 <danb_lap> ?

14:33:52 <db-scribe> nobody implemented it AFAIK

14:47:39 <db-scribe> lots more discussion on DTs

14:51:23 <danb_lap> trying an example: <owlv3:Property rdf:about="http://www.w3.org/TR/REC-rdf-syntax/2003-owl-3.0#logEntails rdfs:label="logical entailment"/>

14:51:55 <db-scribe> measuring how webont would see various solns; not too bad

14:52:06 <db-scribe> measuring how would affect getting docs passed

14:52:14 <db-scribe> impact on syntax

14:54:10 <danb_lap> trying an example: <owlv3:Property rdf:about="http://www.w3.org/TR/REC-rdf-syntax/http://www.oasis.org/2003-owl-4.0#logEntails rdfs:label="logical entailment"/>

14:58:27 <db-scribe> -- coffee

15:20:48 <db-scribe> resumes

15:20:50 <db-scribe> summarizing

15:20:58 <db-scribe> sees 5 as having signficiant impact on webont process

15:21:02 <db-scribe> and extra work for us

15:21:07 <db-scribe> so sees 1 as more appropriate

15:21:11 <db-scribe> and suggest we go for it

15:23:29 <db-scribe> PatrickS outlines a possible problem

15:23:34 <db-scribe> (not really capturing it)

15:24:04 <db-scribe> --

15:24:16 <db-scribe> foo rdfs:subPropertyOf dark:foo

15:24:20 <db-scribe> a foo b .

15:24:21 <db-scribe> entails

15:24:26 <db-scribe> a dark:foo b .

15:24:27 <db-scribe> --

15:26:34 <db-scribe> PatH - valid rdfs, owl reasoning

15:26:49 <db-scribe> jjc - paradox keeps coming is up is onPropertyOf

15:27:05 <db-scribe> peter doesn't want to have to do semantic reasoning; is purely syntactic

15:31:30 <db-scribe> brian: seems ok to continue with #1

15:31:37 <db-scribe> danbri: issue with defining

15:46:10 <db-scribe> more discussion of 5 and nested subgraphs being more research

15:46:21 <db-scribe> and using embedded rdf/xml in the rdf graph as an xml literal

15:47:19 <gk> My thoughts on nested graphs: http://www.ninebynine.org/RDFNotes/UsingContextsWithRDF.html

15:47:26 <db-scribe> but big change to model, ntriples, recursive defns

15:50:48 <db-scribe> brian declares consensus on proposal #1

15:51:24 <db-scribe> ACTION brian: issue rdfms-contexts goes on postponed to next wg list

15:54:44 <db-scribe> proposal 1 is: we will define a computable function on urirefs witll return one of two values - light or dark

15:58:55 <db-scribe> ACTION guha write a proposal for the isdark(uriref) function

15:59:11 <db-scribe> - discussion of a hash function to do this

16:00:43 <db-scribe> ACTION PatH: update the model theory assuming such a function is defined

16:01:39 <db-scribe> ACTION Guha; also consider impact on test cases

16:01:55 <db-scribe> danbri suggests a class in rdfs for these things

16:02:38 <db-scribe> ACTION danbri: proposal a class for rdfs for these dark property things

16:07:21 <db-scribe> PatrickS points out http://www-nrc.nokia.com/sw/draft-pstickler-uri-01.html

16:07:31 <db-scribe> which is an i-d f or URI reification

16:10:10 <db-scribe> ACTION: PatH to draft the dark paragraph for primer

16:10:43 <db-scribe> ACTIOn dave to ensure graph syntax adequately describes darkness

16:11:57 <db-scribe> --

16:13:31 <db-scribe> guha asks to add syntax for XML data types

16:14:12 <db-scribe> <rdf:Description><age xsi:type="xsd:integer">10</age></rdf:Description>

16:14:19 <db-scribe> which would generate triple

16:14:49 <db-scribe> _a: age _b:

16:15:00 <db-scribe> _b: rdf:type xsd:integer

16:15:06 <db-scribe> _b: rdf:value "10".

16:15:10 <db-scribe> -

16:15:56 <db-scribe> PatrickS says should use rdf:type rather xsi:type

16:16:04 <db-scribe> but graph is something proposed ages ago

16:16:49 <db-scribe> guha - reason for doing this way is to make the XML look the same as other uses of XSD in XML

16:16:54 <db-scribe> so that learning curve is 0

16:19:31 <db-scribe> guha - "let's can the cannes proposal - it has a whole can of worms in it"

16:19:44 <db-scribe> patricks - why do literals have global meaning?

16:20:21 <db-scribe> - they have contextual meaning, else why not make them urirefs>

16:22:32 <db-scribe> MEETING CLOSED