RDF Core Working Group IRC logs for 2002-02-25 | |
These are the logs from the
RDF Core Working Group
IRC chat.
Dave Beckett
- [05:21:07] AaronSw
- AaronSw has joined #rdfcore
- [07:07:56] bwm
- bwm has joined #rdfcore
- [07:08:04] AaronSw
- g'morning
- [07:09:22] bwm
- You're up early
- [07:09:26] bwm
- G'morning
- [07:09:29] AaronSw
- no, up late.
- [07:09:35] AaronSw
- :-/
- [07:09:41] bwm
- just finding out we are natted and ipsec does not work
- [07:09:54] AaronSw
- ugh
- [07:09:59] bwm
- i'm off to join the registration line
- [07:10:03] bwm
- back in mo'
- [07:10:06] AaronSw
- c'ya
- [07:14:19]
- * AaronSw wanders off
- [07:14:51] AaronSw
- need some sleep. laters
- [07:18:57] AaronSw
- [BTW, did you get my email bwm?]
- [07:21:28] AaronSw
- i'd appreciate it if we discussed rdfms-fragments first on the telecon, because I have to leave early
- [07:23:20] bwm
- bwm has quit
- [07:24:58] bwm
- bwm has joined #rdfcore
- [07:25:23] AaronSw
- welcome back
- [07:25:23] bwm
- bwm has quit
- [07:26:03] bwm
- bwm has joined #rdfcore
- [07:26:06] AaronSw
- welcome back
- [07:31:31] bwm
- in the room now
- [07:31:40] AaronSw
- did you get my meail? i'd appreciate it if we discussed rdfms-fragments first on the telecon, because I have to leave early
- [07:31:42] bwm
- an i'm live
- [07:31:49] AaronSw
- indeed
- [07:32:03] bwm
- Didn't get your mail - remind me at the time - but sure
- [07:32:13] AaronSw
- great
- [07:35:48]
- * AaronSw is really off now
- [07:42:40] bwm
- am I still connected
- [07:47:18] mdean_
- mdean_ has joined #RDFCore
- [07:47:19] bwm
- bwm has quit
- [07:47:49] mdean_
- hi aaron -- wireless connections are a little flaky here, so we may come and go
- [07:52:12] mdean_
- meeting is starting
- [07:52:25] mdean_
- no great motivational speech
- [07:52:43] mdean_
- em: the restaurant around the corner has vegemite ...
- [07:53:06] mdean_
- brian: welcome
- [07:54:30] mdean_
- libby miller of ILRT is observing
- [07:54:46] mdean_
- objectives on agenda -- goal to get issues down to 0
- [07:55:02] mdean_
- brian: need to think about getting to last call and beyond
- [07:57:49] libby
- libby has joined #rdfcore
- [07:57:49] mdean_
- mdean_ has quit
- [07:58:27] mdean_
- mdean_ has joined #RDFCore
- [07:59:06] mdean_
- original call -- last call October 2001
- [07:59:33] mdean_
- issue: rdfms-not-id-and-resource-attr
- [08:00:21] mdean_
- jeremy: rdf:ID on property elements always used for reification
- [08:00:36] mdean_
- dajobe: seconded
- [08:07:26] mdean_
- examples being entered to http://www.w3.org/2001/sw/RDFCore/20020225-f2f/session1.html
- [08:07:58] mdean_
- dajobe: this will require change to test case 5 and probably others
- [08:08:17] mdean_
- ACTION: Jeremy formalize test case
- [08:08:29] mdean_
- ACTION: Jan update test cases document
- [08:08:51] mdean_
- with this test case
- [08:09:17] mdean_
- review other test cases for side effects (empty-property-elements 005)
- [08:09:51] mdean_
- 23 issues to go
- [08:10:24] mdean_
- issue: rdfms-nested-bagIDs
- [08:13:00] mdean_
- paste example into W3C web site -- ARP does this right
- [08:19:26] libby
- dave's mailt to rdfcore: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Jan/0113.html
- [08:19:26] mdean_
- mdean_ has quit
- [08:19:51] libby
- is everyone happy with this :
- [08:20:51] libby
- a bag id reifies the property attributes of the same elemnt of the bag id, the typed node, and the property elements that are direct children of that node and nothing else (jeremy)
- [08:21:02] mdean_
- mdean_ has joined #RDFCore
- [08:21:34] libby
- in particular a prop element that itself has prop attributes, those attributes are not part of the bag
- [08:21:34] mdean_
- mdean_ has quit
- [08:22:27] libby
- 'shallow as it can be'
- [08:23:08] libby
- (says jeremy)
- [08:23:24] libby
- jos: motivation?
- [08:23:39] libby
- dave: simplest answer
- [08:24:08] libby
- jos: in a sense not logical
- [08:24:25] libby
- ...reify everything...
- [08:24:34] libby
- jos has a proble with this
- [08:24:53] libby
- ...a bag of statement is a bag of statements, all of them
- [08:26:14] libby
- brian: look at orginal example
- [08:26:23] libby
- ..this generates 2 bags
- [08:26:51] mdean_
- mdean_ has joined #RDFCore
- [08:26:59] libby
- (differs a bit from dave's example)
- [08:27:08] libby
- bag1 contains st1
- [08:27:19] libby
- bag2 constains st2
- [08:27:25] libby
- no typed node in this example
- [08:27:49] libby
- [[
- [08:28:13] libby
- <rdf:Description about="a" bagID="bag1">
- [08:28:29] libby
- <some:prop rdf:ID="st1">
- [08:28:38] libby
- ... see doc
- [08:28:41] libby
- approved!
- [08:29:09] libby
- actions:
- [08:29:15] libby
- dave to update the document
- [08:29:52] libby
- brian to review testcases tomorrow in detail
- [08:30:36] mdean_
- wireless connection seems much better now -- antenna has been moved
- [08:31:24] JosD
- JosD has joined #rdfcore
- [08:32:59] libby
- jjc generate more testcases
- [08:33:06] em-cann
- em-cann has joined #rdfcore
- [08:33:14] libby
- brian: closed issue, and also the provious one
- [08:33:29] libby
- 2 down!
- [08:33:57] mdean_
- next issue: rdfms-rdf-names-use
- [08:34:58] em-cann
- ping
- [08:36:38] libby
- brian: waiting to complete the test cases
- [08:36:48] libby
- ...action on dave...
- [08:37:01] libby
- dave: infinite!
- [08:39:27] libby
- brain: do we restrict rdf/rdfs names which are not syntactic? no
- [08:39:46] libby
- jjc: rdf:foo e.g? things that arent currently used?
- [08:40:01] libby
- ..but in the rdf ns
- [08:40:20] libby
- dave: resrve these
- [08:40:26] libby
- brain disagrees
- [08:40:34] libby
- s/brain/brian
- [08:40:48] libby
- ...existing parsers will barf
- [08:41:39] libby
- jjc: parser issues warning then treats it as normal. can still call it illegal
- [08:42:19] libby
- gk: any definition nof meaning is reserved for a future wg
- [08:42:28] libby
- brain: what triples?
- [08:42:49] libby
- brian: I suggest allowed
- [08:43:10] libby
- ps: parsers dont have to die even if an error
- [08:44:10] libby
- gk: we dont want to commit its an error for all time. 'undefined' category for test cases?
- [08:44:22] libby
- dave: error
- [08:44:27] libby
- jjc: error
- [08:44:30] libby
- ps: warning
- [08:44:41] libby
- maryn: error
- [08:44:45] libby
- gk:warning
- [08:44:55] libby
- danc: error but dont care
- [08:44:59]
- * libby lost
- [08:45:33] libby
- eric: generate the triple
- [08:46:30] libby
- brain: m&S says rdf ns is reserved
- [08:46:37] libby
- danc: not important enough
- [08:46:53] libby
- brian: answer: generats the triple and should generate the warning
- [08:47:02] libby
- issue closed!
- [08:47:08] libby
- countem, 3!
- [08:47:29] libby
- action: daveb update syntax doc
- [08:47:43] libby
- jang ensure testcases created andf added to testcases doc
- [08:48:04] libby
- jjc: testcase manifest needs to eb able to say this testcase geneartes a warning
- [08:48:18] libby
- ...currently doesnt do this
- [08:48:46] mdean_
- issue: rdfms-xml-base
- [08:49:21] libby
- jjc: needs to go to TAG. same doc references do not get proceesed wrt the base uri
- [08:49:49] em-cann
- em-cann has quit
- [08:50:08] em-cann
- em-cann has joined #rdfcore
- [08:51:07] libby
- lots of test cases available; they doesnt use rfc 2396 though; but is we use that rfc, rdf:ID becomes counter-intuitive
- [08:53:29] libby
- see http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0516.html
- [08:53:29] em-cann
- em-cann has quit
- [08:53:48] libby
- test 8
- [08:55:00] libby
- jjc: not our problem, its a prob w 2396, maybe a TAG problem
- [08:56:18] libby
- brian: proposal is are the testcases with the same document erference acceptable
- [08:56:40] libby
- 3 people have implementations which are affected
- [08:56:57] libby
- ...or who want to, especially mike, who thinks this implementation makes sense
- [08:57:09] libby
- danc: we nede to ask the powers that be at last call
- [08:57:41] libby
- brian: approved!
- [08:58:12] libby
- actions: daveb update syntax doc;
- [08:58:23] libby
- jang update test cases
- [08:59:01] libby
- danc: we need to ask the editors of 2396 if this is ok: action danc
- [08:59:38] libby
- danc: syntax needs to express teh algorithm used
- [08:59:40] gk
- gk has joined #rdfcore
- [08:59:45] libby
- jjc: to write it up
- [09:00:46] libby
- danc: thinks this strategy will fail
- [09:01:50] libby
- em: can get different conclusions of rfc 2396
- [09:01:50] libby
- 'reasonable peopel can disagree'
- [09:02:32] libby
- em: the best approach might not be to emphasise the differences
- [09:02:54] DanC
- DanC has joined #rdfcore
- [09:02:55] libby
- danc: in this context the exception doesnt apply
- [09:03:15]
- * DanC wonders if this channel is to be used for records
- [09:03:25] libby
- I've been making notes
- [09:03:55] libby
- ..here
- [09:04:16] libby
- em: suggests daveb write this, and persuade jjc to help; danc talk to rfc editiors
- [09:05:20] libby
- brian: action jjc write up the case that 2396 context is different to browsing
- [09:05:36] libby
- em: multiple codebases woudl help - mike?
- [09:05:58] libby
- mike uses jena/arp
- [09:06:36] libby
- not mutiple codebases
- [09:06:44] libby
- daveb will imleemnt this in raptor
- [09:09:22] libby
- jjc: we've solved the first of 4 issues in http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0516.html
- [09:10:24] libby
- jjc: problem: base url has no end slash, fragment doesnt have a start slash. jjc reads 2396 to say no slash separator
- [09:11:00] libby
- ...minor bug rather than change in intent
- [09:11:13] libby
- danc: recommends the same action as previously
- [09:11:27] libby
- gk: ask tthem if the slash is what they really meant
- [09:11:42] libby
- jjc: hold approval of testcase till checked w them?
- [09:12:16] libby
- brian: approve testcase, action to review it with the 2396 editors. danc will do this latter
- [09:12:55] libby
- testcases are sufficient
- [09:13:01]
- * DanC hears that Larry left; wonders where he went
- [09:13:06] DanC
- logger_3, pointer?
- [09:13:06] DanC
- See http://ilrt.org/discovery/chatlogs/rdfcore/2002-02-25#T09-13-06
- [09:13:13] libby
- testcase is approved and issue closed (brian)
- [09:13:28] libby
- testcase 11, that is from http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0516.html
- [09:14:31] libby
- jjc next issue from the email: testcase 14
- [09:15:34] libby
- ...2 occurances of rdf:ID="frag" in same doc is an error; propose if xml:base differ then allow it
- [09:15:52] libby
- dave: drop restriction that fragids be checked for uniqueness?
- [09:16:06] libby
- mike finds duplicate id detection v useful
- [09:16:35] libby
- jjc: kinda useful for reification
- [09:16:43] libby
- danc: thinks shoudl work like this
- [09:17:04] libby
- jjc: check for dups after extending the base
- [09:17:15] libby
- :approved!
- [09:17:38] libby
- drop duplicate detection altogetther for efficincy sake?
- [09:18:29] libby
- jjc: people could use about in a v long doc
- [09:19:11] libby
- brian: keep id frag checking
- [09:19:15] libby
- no actions
- [09:19:58] libby
- danc duplicate ids is an error means not an rdf document
- [09:20:06] libby
- mike like sthe warning
- [09:20:19] libby
- jjc: arp will barf if strict flag
- [09:20:24] libby
- 4th base issue
- [09:21:01] libby
- jjc: several error case: error1,2,3
- [09:21:30] libby
- ...1 really is an error: 'mailto uri - no way to resolve this base
- [09:21:36] libby
- ...not an rdf document
- [09:21:52] libby
- em disagrees
- [09:22:11] libby
- jjc: no, no algorithm to generate this
- [09:22:27] libby
- danc: cant merge the path here, because mailto no path. agree error
- [09:23:35] libby
- jjc :error case 2: a reasonable algorithm could generate this, but jjc thinks an error
- [09:24:16] libby
- ...foo is a same document reference not a relative uri. these definitely different. rfc expliictly says
- [09:25:25] libby
- case 2 is dropped, action on jjc to create a positive testcase
- [09:25:34] libby
- error case 3:
- [09:26:17] libby
- same doc reference with a mailto uri. shoudl go the same way as error case 2
- [09:26:29] libby
- action jjc to craete positive case
- [09:27:23] libby
- brian: do rest of those testcases tomorrow morning
- [09:27:50] libby
- jjc: that's it with xml:base
- [09:28:08] libby
- jjc found the xml:base issues. we are stil on for doing xml:base.
- [09:28:16] libby
- only 20 issues left!
- [09:28:22] libby
- groups breaks for coffee
- [09:29:08] libby
- libby has quit
- [09:57:32] libby
- libby has joined #rdfcore
- [09:58:11] libby
- libby has quit
- [10:00:04] libby
- libby has joined #rdfcore
- [10:04:34] libby
- rdfms-seq-representation is done already
- [10:04:34] libby
-
- [10:04:34] libby
- dave: we resolved on 11 jan not to change the rpoperty
- [10:04:34] libby
- brain: but what it means needs to be clarified in the context of datatypes
- [10:05:04] libby
- ...a specific understanding of value for dts will break existing code
- [10:05:51] libby
- ...proposes: no model theory for it, but its meaning is entirely app-dependent. suggested styles of using it
- [10:05:55] libby
- jjc seconded
- [10:06:23] libby
- ps would like it to be that this does not preclude it from being used in a datatyping idiom
- [10:08:14] libby
- danc would not be happy w putting 1.0 examples in here. woudl like real uses
- [10:08:14] JosD
- JosD has quit
- [10:08:22] libby
- dave: dublin core uses it
- [10:08:40] libby
- danc: does dc usage agree w 1.0 spec?
- [10:09:29] libby
- em: useful for UI
- [10:09:46] libby
- danc: who is going to write this use guide?
- [10:09:54] libby
- em/frank
- [10:10:12] libby
- - frank is primer editor
- [10:10:38] libby
- brain: propose the wg resolve:
- [10:10:53] libby
- - rdf:value of a property defined in the rdf ns
- [10:11:05] JosD
- JosD has joined #rdfcore
- [10:11:34] libby
- that no model theory semantics is defined spceifcally for it
- [10:12:17] libby
- action em to ensure primer describes appropriate use of rdf:value
- [10:12:38] libby
- danc: likely that phayes will likely want to say soemthing about it
- [10:13:55] libby
- - model theory must state that rdf:value is a property
- [10:15:29] libby
- everyone is happy
- [10:15:48] libby
- action - bwm ensures phayes ensures in mt
- [10:16:29] libby
- rdfms-editorial
- [10:17:59] libby
- danc: are we taking rdfs 1.0 forward or not..?
- [10:17:59] libby
- jjc: propose no v2 of rdf m&s? supceceded by testcase docs and primer, syntax doc and teh model theory
- [10:18:16] libby
- brian: proposed
- [10:19:05] libby
- danc primer discusses the other docs
- [10:19:40] libby
- approved.
- [10:20:14] libby
- issue continues: rdfms-editorial: typos, unclear language etc
- [10:20:44] libby
- brian: propose that those issues now not relevant and we close the issue
- [10:21:22] libby
- ...the editors of teh docs would look at the detil of the issues so dont make the same mistakes again
- [10:21:29] libby
- danc: need an errata page
- [10:22:44] libby
- action em: update M&S errata to say that m&s will be superceded by the new documents
- [10:23:21] libby
- closed.
- [10:24:25] libby
- next issue: rdfms-fragments
- [10:26:22] libby
- gk: if it is an issue, fragment ids dont behave in quite the same way in rdf as with other web retrieval dodads
- [10:27:32] libby
- gk: orginal email:http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0559.html
- [10:27:49] libby
- ...if you ahve different mime types, fragment ids are used in different ways
- [10:28:13] libby
- ...proposal was to say that rdf uses the fragment identifier in its own way
- [10:28:50] libby
- dave: my rdf code doesnt do retrivel
- [10:29:34] libby
- brian: wording in spec or what peple expect
- [10:30:02] libby
- ...2 cases fragment which is a class, for example. w mimetype we can contriol this in rdf
- [10:30:33] libby
- ...second case: chapter 2 written by mike, etc
- [10:30:56] libby
- ...semantics could mean a diffeent thing that what mean by it
- [10:31:41] libby
- danc: trying to create a testcase for it
- [10:32:30] libby
- ...suppose wanted to point to second description in an rdf doc using xlink
- [10:32:50] libby
- ...trying to write an rdf doc w the same uri erference, and see whether it works
- [10:33:31] libby
- danc: timbl I think thinks that one points to the think and one points to the reference to the thing - difference
- [10:33:55] libby
- gk: maybe if no testcases not important. doe sit work w uris with no hash fragments?
- [10:34:08] libby
- http://www.ietf.org/rfc/rfc2396.txt 4.1 fragment identifier
- [10:34:48] libby
- ...'resources in rdf dont correpond to anything called resource son the wider web if no #...?
- [10:35:35] libby
- graham's proposal :http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0559.html
- [10:37:14] libby
- aaron has a proposal: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0391.html
- [10:37:33] libby
- danc: doesnt like it; jjc, too big a change. wide disagreement
- [10:37:47] libby
- gk: use of hash fragments is too well established
- [10:38:15] libby
- brian: as aaron: changing eartly is better if have to change
- [10:38:24] libby
- gk: inelegant, but dont have to change
- [10:39:13] libby
- brian: has come up on mailing list
- [10:41:01] libby
- gk thinks has a testcase
- [10:41:01] libby
- also so does jjc
- [10:41:19] libby
- danc: cant serve rdf as text/xml, cos then fragids mean what they d in the rfc
- [10:41:27] libby
- (says timbl)
- [10:41:44] libby
- ditton the cuurent rdf mime proposal
- [10:42:37] libby
- gk: a doc in diff mime formats, and we want to make statements about fragments of the document independently of its mimetypes
- [10:43:30] libby
- jjc: similar test case, all text xml docs of diff sizes after content negotiation
- [10:43:52] libby
- ....23rd element is different things in each document
- [10:44:01] libby
- ...ill-defined because document is illdefined
- [10:44:08] libby
- danc: just dont do that
- [10:45:23] libby
- gk just want to flag the problem
- [10:46:43] libby
- ....arguing....
- [10:47:52] libby
- brian: first case is repesentation idependent; second case (jjc, #23) is not
- [10:49:48] libby
- gk: can make true rdf statements about fragment ids, which makes no guarantss about what you get if you do a webretrieval
- [10:49:48] JosD
- JosD has quit
- [10:49:56] libby
- danc: that makes it false
- [10:50:59] libby
- jjc: possible workds thing: content negotiation means that your view of web is diferrent from mine. trye for you not necessarily true for me
- [10:52:47] libby
- em: heads up and move on? not changing documents
- [10:53:49] libby
- brian: does anyone support aaron's proposal?
- [10:53:52] libby
- no..
- [10:55:01] libby
- gk: esssence of the proposal is first 2 paras of email
- [10:56:06] libby
- proposed: that fragmenmt identifiers when used with rdf are treated as simple extenion of the URI to which they apply, without having an interpretatiuon that is in any way dependant on the context in which they appear
- [10:56:08] gk
- Fragment identifiers and truth: different people may view documents-with fragment ids and get different values; thus the truth about an RDF statement using a frag id may not be universal to all Web users
- [10:57:25] libby
- dave: change 'context'
- [10:58:09] libby
- gk: something for the primer maybe?
- [10:59:12] libby
- brian: proposed: RDF uses URIs with fragment IDs to identify resources
- [11:00:56] libby
- action gk to draft somethign for the primer?
- [11:02:43] libby
- danc: this issue has been accepted by the TAG
- [11:03:19] JosD
- JosD has joined #rdfcore
- [11:03:21] libby
- danc: they've accepted qname-nm-mappings
- [11:03:29] libby
- ...soirry, it's not on the TAG list
- [11:04:06] libby
- action danc: highlight this issue with TAG
- [11:04:15] libby
- - done for now on thios issue
- [11:06:05] libby
- -donr all issues for today except those on telecon, and datatyping after lunch
- [11:06:10] libby
- lets have lunch.
- [11:06:15] libby
- (says most people)
- [11:07:28] libby
- quick meeting re tomorrow
- [11:08:35] libby
- gk talked to micha last night
- [11:09:14] libby
- ...problem of matching literals in different xml:langs
- [11:11:32] gk
- Problems misha raised (with JJC's proposal) were: (1) language of parts as well as whole, (2) language matching rules not so clear-cut, (3) ~fuzzy report~ issues with capturing subtle linguistic contexts
- [11:13:04] gk
- Sort-of example of that discussion: document --dc:title--> "The many meanings of 'chat'"
- [11:13:31] gk
- document --ex:subjectMatter--> "chat" .
- [11:14:56] libby
- ps: can we make it brutally clear that parsetype literal is in fact parsertype xml?
- [11:16:41] libby
- danc: reseralizing this bit of xml loses the xml part - get < etc. then the bit in the xml stops being italian
- [11:16:54] libby
- ...and turns into english (jjc)
- [11:18:15] libby
- danc could have xml(...) in ntriple
- [11:19:45] libby
- brian: aaron wanted to record the parsetype flag in the ntriple, preferably as a uri
- [11:20:50] libby
- ...otherwise we lose information
- [11:21:05] libby
- ...arguing....
- [11:22:33] libby
- brain suggests putting the thing in the parsetype inverted commas as the first part of teh ntriple
- [11:22:40] libby
- ...later might be a uri
- [11:22:55] libby
- danc: people use literal in parsetype as a qname
- [11:23:56] libby
- em: consensus about somekind of indication
- [11:26:44] libby
- ..arguing...
- [11:27:01] libby
- danc: nothing was promised to make parsetype an extensible mechanism
- [11:28:15] libby
- daml uses it; others use it as extensible machanism
- [11:28:28] libby
- brian: proposal: uri
- [11:29:13] libby
- ...undecided for now
- [11:29:49] libby
- back 13.30
- [11:29:54] libby
- libby has quit
- [11:34:13] gk
- gk has quit
- [11:50:47] DanC
- DanC has quit
- [11:50:47] JosD
- JosD has quit
- [11:52:54] mdean_
- mdean_ has quit
- [12:24:44] DanC
- DanC has joined #rdfcore
- [12:37:43] libby
- libby has joined #rdfcore
- [12:39:25] gk
- gk has joined #rdfcore
- [12:40:14] gk
- I18N discussion continues...
- [12:40:37] gk
- PatStickler promises not to say anything of allowed to lurk!
- [12:40:53] gk
- in tomorrows meet with I18N group
- [12:41:40] gk
- Trying to clear up I18N issues *before* we go to last call.
- [12:42:16] gk
- Objectives of tomorrow#'s meeting?
- [12:43:01] gk
- Objectives of tomorrow#'s meeting?
- [12:43:06] gk
- jjc: 3 issues from charmod, identify what we're trying to do in broad terms, what to do with N-triples.
- [12:44:01] gk
- assues arise from our reading of charmod, want to check with I18N that our proposed resolutions will meet their requirements.
- [12:44:30] gk
- Also, they may have other issues for us, and we need to agree the boundaries on issues we can reasonably consider.
- [12:45:27] danbri_
- danbri_ has joined #rdfcore
- [12:45:36] gk
- E.g. literal matching proposed match only if string and lang tags are ==. But in linguistic terms, that's too, er, literal. How muchj of this is reasonably our problem to solve?
- [12:45:51]
- * gk Hi danbri
- [12:45:56]
- * danbri_ sneaks in through the irc backdoor, wonders how it's going
- [12:46:04] libby
- ahoyhoy danbri
- [12:46:12]
- * gk good, so far. Good number of issues settled.
- [12:47:03] gk
- jjc: we're trying to build a sound basis for RDF; that needs we well-defined concept of literal equality.
- [12:47:03] danbri_
- excellent news! so sorry not to be there... I'll lurk here, see if I can absorb some of the atmosphere...
- [12:47:23] libby
- I'll take some photos to help you
- [12:47:32] gk
- Beyond that it's less cear what we need to address in this WG.
- [12:48:19] gk
- PatS: queries may have deeper linguistic qualifications (e.g. recognizing that unequal literals to refer to some common idea).
- [12:49:11] gk
- jjc: Note - our spec is NOT RDF query.
- [12:49:25] gk
- (GK nods vigorously)
- [12:49:29] mdean_
- mdean_ has joined #rdfcore
- [12:50:38] gk
- jjc: Another issue... parsetype-literal ... sensitivity to language used inside a complex literal.
- [12:50:55] gk
- This is Hard Work.
- [12:51:27] gk
- Probably not entirely our problem.
- [12:52:39] gk
- GK expresses concern that we have parallel structuring mechanisms
- [12:52:53] gk
- PatS: extra triples suggested, but shot down
- [12:52:57] JosD
- JosD has joined #rdfcore
- [12:53:50] gk
- Who is going: to the I18N meeting? JJC, DaeB, Martyn, bwm, GK
- [12:54:26] gk
- ====== Datatypes ========
- [12:55:35] gk
- PatS has produced a new proposal, since last Friday. Need to catch up.
- [12:55:58] JosD
- http://www.coginst.uwf.edu/users/phayes/simpledatatype23-02-2002.html
- [12:56:12] gk
- bwm: want to bring us all up to the same point, somehow, if we can.
- [12:57:27] gk
- bwm: taken a long time, but have produced a common proposal (one of several)
- [12:58:30] gk
- Last Friday we talked around it; specific questions: do we want the idion known as S-B (as syntax)? YES.
- [12:59:15] gk
- If we say in S-B syntax: <Mary> <age> "10" . Does "10" denote a number or a string? A: string.
- [12:59:27] gk
- Do we want to keep the so-called doubled idion? A: No
- [12:59:51] gk
- Do we want to keep the so-called triplet idion? A: YES
- [13:00:11] gk
- ---- all thje abovre from last Fridays' telecon ---
- [13:00:38] gk
- New proposal from PatS, who was unhappy (at the time) that the requirements were in conflict.
- [13:02:56] gk
- DanC: wants to check real use-cases from real applications against proposals. But is appy with answers from Friday.
- [13:05:53] gk
- (Discussion about what point to move forward from...)
- [13:06:35] gk
- jjc (aka Unhappy of bristol) will put up his entailment case that he wants discussed.
- [13:07:14] DanC
- unhappy msg: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0635.html
- [13:09:59] gk
- jjc: wants to entail equivalence across both forms of idiom
- [13:17:31] gk
- Trying to show how inference can be drawn ... looking at PatrickS email from Sunday: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0642.html
- [13:34:15] libby
- ...arguing, shouting, jumping up....
- [13:36:50] libby
- jos: you could have 2 different ranges, one a drange then patrickS's proposal is ok
- [13:37:20] libby
- danc: end up with 2 property names for similar properties
- [13:37:43] libby
- brian: jjc's entailments will work if we have range and drange, we think
- [13:38:09] libby
- jjc will then take objection off the table
- [13:38:26] libby
- brain: then the inelegance is the 2 properties for similar things
- [13:38:47] libby
- patrickS thinks he has the solution why need only one
- [13:39:06] libby
- ps: classes contains members
- [13:40:59] libby
- ...if a datatype name is only the value, we need to have somthuing to denote the lexical space; if some people want one and some want the other, cant use range property to denote both
- [13:41:21] gk
- Here's a rendering of what I think Patrick is trying to suggest...
- [13:41:23] gk
- # datatype triple idiom:
- [13:41:23] gk
- {
- [13:41:23] gk
- :x :p [ :d :l ] . // Tom ageA [ integer "10" ]
- [13:41:23] gk
- :d rdf:type rdfd:Datatype . // integer a Datatype
- [13:41:23] gk
- # :l rdf:type rdfs:Literal .
- [13:41:24] gk
- }
- [13:41:26] gk
- log:implies
- [13:41:28] gk
- {
- [13:41:30] gk
- :x :p [ rdf:type rdfd:DatatypeValue ; // Tom ageA [ a Datatype Value ;
- [13:41:32] gk
- rdfd:datatype :d ; // datatype integer ;
- [13:41:34] gk
- rdfd:lexicalForm :l ] . // lexicalform "10" ] .
- [13:41:36] gk
- } .
- [13:41:38] gk
- # inline idiom:
- [13:41:40] gk
- {
- [13:41:42] gk
- :x :p :l . // Jenny ageB "10
- [13:41:49] gk
- :p rdfs:range :d . // ageB rdfs:drange integer .
- [13:41:49] gk
- :d rdf:type rdfd:Datatype . // integer a datatype
- [13:41:49] gk
- # :l rdf:type rdfs:Literal .
- [13:41:50] gk
- }
- [13:41:52] gk
- log:implies
- [13:41:54] gk
- {
- [13:41:56] gk
- :x :p [ rdf:type rdfd:DatatypeValue ; // Jenny ageB [ a DataTypeValue ;
- [13:41:58] gk
- rdfd:datatype :d ; // datatype integer ;
- [13:42:00] gk
- rdfd:lexicalForm :l ] . // lexicalForm "10" ] .
- [13:42:02] gk
- } .
- [13:42:23] libby
- ....rules disambibuate which is which
- [13:43:24] libby
- jjc: thinks might work apat form some baroque datatypes
- [13:44:41] libby
- jjc: "monday -> "monday"
- [13:44:49] libby
- .. "lundi" -> "monday"
- [13:45:10] libby
- ..."tuesday" -> monday (in some language like klingon., say)
- [13:45:49] libby
- ps: values do not have canonical represtation in teh model theory
- [13:45:57] libby
- ...people struggling a bit...
- [13:46:42] libby
- jjc: is this it:? likelihood with same lirteral on both sides is very unlikely?
- [13:47:03] libby
- brain: do we care about a corne case like that
- [13:47:34] libby
- jos: I care, because pat had good examples of why it might go wrong
- [13:48:07] libby
- danc: some consensus earlier (re 2 properties?) move on?
- [13:48:27] libby
- ps think this proble is crucial to understanding the problem
- [13:48:44]
- * libby not sure about this "below the line" stuff
- [13:48:55] libby
- "to the left of the line"?
- [13:49:09] gk
- I think ":below the line" is a way to refer things in the semantic domain rather than the syntactic domain
- [13:49:26] libby
- yeah, thought somethign like that
- [13:49:45] libby
- ericp does pat understand ps's problem?
- [13:49:47] libby
- ps - no
- [13:50:01] gk
- The "idioms" (A,B,etc.) are all purely syntactic. The specific denotations of values involved are semantic.
- [13:52:07] libby
- gk might have a better example...
- [13:52:51] libby
- brian: dancs point was that we have agreement, but we arguing over the detail of 1 or 2 properties
- [13:53:47] libby
- jjc: partially confinced: ps's example so ugly must be wrong but cant see how!
- [13:54:12] libby
- ditto danc
- [13:54:29] libby
- brian thinks breaks current model theory, requires lv to be a subset of ir
- [13:56:11] libby
- ps thinks 'for the users' one prop is better
- [13:56:25]
- * AaronSw wakes up, waves quietly
- [13:56:28] libby
- brian: all agree that _if_ we dont need drange, we dont want it
- [13:56:32] libby
- yep
- [13:57:21] libby
- ps would need to 2 flahours of schema, one for each idiom
- [13:58:20] libby
- gk: 2 discssions: range, drange; and ageA and ageB
- [14:03:22] gk
- State of discussion: jeremy is now "less unhappy" of bristol, but still not overjoyed.
- [14:03:33] libby
- :)
- [14:04:59] gk
- Progress: we think it *may* be possible to unify (entail between) S-A/S-B approaches with some additional rules along the lines suggested by Patrick
- [14:06:36] gk
- Jeremy is not unhappy about having two properties for the diferent idioms
- [14:07:28] gk
- Jos has proved that entailment is possible if different property names (ageA/ageB) are used.
- [14:08:33] gk
- Agreed that entailment between idioms must be possible.
- [14:10:06] libby
- jos cannot make jjc's entiaolment work with only one property
- [14:10:15] gk
- Jos has been unable so far to make entailment work with one property, without causing a conflict.
- [14:11:20] libby
- jjc the rules will be in the model theory ? yes
- [14:11:37] libby
- danc: ps: just one age propertry
- [14:11:54] gk
- Patrick thunks we can get by with just one age property, by alowing properties to have lexical and value space in range.
- [14:12:19] gk
- (i.e. datatype class has both lexical and value spaces as members)
- [14:13:12] libby
- jjc: near-unanimity that ps's suggestion is ugly ;)
- [14:13:29] libby
- danc: also ageA and ageB is ugly
- [14:13:32] gk
- However, there are different view about which of the possible solutions is uglier.
- [14:14:35] gk
- Also, Patrick thinks (with some support) we can do without drange.
- [14:14:55] libby
- danc: so costly to explain to users
- [14:16:25] libby
- martyn: intuition is ugly
- [14:16:37] libby
- jos: I would say mathematically inconsistent
- [14:16:45] libby
- bad patrick!
- [14:17:22] gk
- Jos says he thinks he will be able to find an inconsistency in Patrick's proposal, but has not yet been able to do so.
- [14:19:11] gk
- (at this stage, we're basically wordsmithing the conclusions on screen.)
- [14:19:25]
- * gk Hi aaron
- [14:19:33]
- * AaronSw waves
- [14:19:41]
- * gk we're just breaking for coffee 15:20 loal time
- [14:21:00] gk
- Back in 40mins? (maybe sooner -- some rambling discussion) Telecon at 16:00 local?
- [14:21:27] AaronSw
- My schedule says telecon in 40 min.
- [14:21:50] AaronSw
- yeah, so 1600
- [14:21:58] libby
- libby has quit
- [14:22:57] gk
- Aim to beback in the room 25mins... people leaving now
- [14:47:09] gk
- Wanders back into room ... DanC and PatrickS going hammer-and-tongs about someting on the whiteboard ...
- [14:49:54] AaronSw
- Heh. hammer-and-tongs?
- [14:50:13] AaronSw
- i asumme that means something like full-out/crazy
- [14:50:46] gk
- Yeah... English expression ;-)
- [14:52:43] bwm
- bwm has joined #rdfcore
- [14:55:23] libby
- libby has joined #rdfcore
- [15:01:36] DanC
- hi, anybody?
- [15:01:41] AaronSw
- hello
- [15:02:23] AaronSw
- <Zakim> SW_RDFCore()10:00AM has now started
- [15:03:03] bwm
- hi - zakim has not called us
- [15:03:11] AaronSw
- what's the number?
- [15:03:15] AaronSw
- <Zakim> +Manola
- [15:03:15] AaronSw
- <Zakim> +Guest P2 7332 - is perhaps PatH?
- [15:03:20] bwm
- Dan is wokring on it
- [15:03:48] DanC
- "Your request at 15:03Z to add TP-Iles-C to the Zakim bridge conference 7332 has been added to the queue."
- [15:04:19] AaronSw
- <Zakim> +TP-Iles-C
- [15:04:40] AaronSw
- <Zakim> -PatH?
- [15:04:41] AaronSw
- <Zakim> -TP-Iles-C
- [15:05:03] AaronSw
- <Zakim> +Guest P2 7332 - is perhaps PatH?
- [15:05:03] AaronSw
- <Zakim> +AaronSw
- [15:05:32] DanC
- hmm... we were there and then left?
- [15:05:45] bwm
- aaron - are you dialed in - please tell pat we are trying
- [15:07:26] AaronSw
- yep, we're making all the decisions ourselves
- [15:07:46] bwm
- cool - we can check if they are the same as ours
- [15:08:15] AaronSw
- We've come up with a new insult: "you bNode!"
- [15:08:57] ircleuser
- ircleuser has joined #rdfcore
- [15:09:38]
- * AaronSw waves to PatH?
- [15:09:47] ircleuser
- hithere
- [15:09:54] DanC
- we're experiencing dialing difficulties.
- [15:10:21] ircleuser
- ircleuser is now known as pathayes
- [15:13:39] bwm
- Hi pat
- [15:13:47] AaronSw
- <Zakim> +TP-Iles-C
- [15:13:55] AaronSw
- you guys sound like a busy signal
- [15:14:02] AaronSw
- <Zakim> -TP-Iles-C
- [15:14:39]
- * gk still problem dialling
- [15:15:14] AaronSw
- <Zakim> +TP-Iles-C
- [15:15:18] AaronSw
- same thing
- [15:15:38] AaronSw
- <Zakim> -TP-Iles-C
- [15:15:49] AaronSw
- you connect and all we hear is bleeeep bleeeep bleeeep bleeeep
- [15:15:51] bwm
- k
- [15:17:07] AaronSw
- we think the problem is that you've identified your telephone by using a fragment uri
- [15:17:07] gk
- I must tone down my language ;-)
- [15:21:57]
- * DanC chacls minutes so far
- [15:24:46] DanC
- I think it's silly that we don't conclude that thinks with the same subj/pred/obj are equal. I'd like to abstain.
- [15:24:54] DanC
- ... if the logisitcs allow.
- [15:25:52] DanC
- I think this is available to the world now. http://www.w3.org/2001/sw/RDFCore/20020225-f2f/20020225-am.html
- [15:27:00] AaronSw
- seems to be
- [15:27:09] gk
- DanC: it's -pm I cannot get to.
- [15:27:15] DanC
- I'm working on it...
- [15:27:44] DanC
- chair: pls see -am notes re fragments issue...
- [15:27:55]
- * DanC wonders who's the scribe
- [15:28:31] DanC
- aaron or pat, confirm you're looking at fragments notes from this morning? (i.e. what the chair is reading?)
- [15:28:36] DanC
- AaronSw: sounds good.
- [15:28:42] AaronSw
- yes
- [15:28:46]
- * danbri_ tries to dial in
- [15:28:52] AaronSw
- "fragments may cause cancer in mice"
- [15:29:11] DanC
- pathayes: s/identifiy/denote/ ? or locate?
- [15:29:39] DanC
- path, you have an action to edit the model theroy spec as appropriate.
- [15:29:57] DanC
- (provided you accept; oops; recorded action was for the chair to ask you to do that)
- [15:30:03] DanC
- nak, pat. say again.
- [15:30:04]
- * danbri_ dissapears from bridge dialout queue; still waiting for phone to ring.
- [15:30:46] DanC
- pathayes: name?
- [15:30:47] gk
- --- fragment ids...
- [15:30:52] gk
- --- fragment ids...
- [15:31:02]
- * gk test
- [15:31:06] DanC
- EricM abstains.
- [15:31:13] AaronSw
- Aaron abstains.
- [15:31:21] DanC
- RESOLVED. (not sure where the real record is)
- [15:31:53]
- * DanC takes scribe duties
- [15:32:48]
- * danbri_ gives up on phone bridge dialout; returns to lurking
- [15:32:53] DanC
- RESOLVED: Proposed: RDF uses URI's with fragment ID's to name resources. this closes rdfms-fragments.
- [15:33:03] DanC
- RESOLVED: RDF uses URI's with fragment ID's to name resources. this closes rdfms-fragments.
- [15:33:11] DanC
- reiterate:
- [15:33:13] DanC
- Action: Graham draft text for the primer, including a "here be dragons" warning
- [15:33:13] DanC
- Action: DanC Highlight this issue with the TAG
- [15:33:23] DanC
- ------- Datatypes
- [15:33:40] gk
- See -pm notes ... dan is working to fix protection
- [15:33:44] DanC
- chair: can you see the -pm notes?
- [15:34:05] AaronSw
- pat, aaron: no
- [15:34:11] DanC
- done.
- [15:34:13] JosD
- http://www.w3.org/2001/sw/RDFCore/20020225-f2f/20020225-pm.html
- [15:34:15] bwm
- C:\User\w3c\WWW\2001\sw\RDFCore\20020225-f2f\20020225-pm.html
- [15:34:42] bwm
- We have agreed that some form of Jeremy's entailment is required (msg 0635 Feb)
- [15:34:42] bwm
- Patrick's entailment suggests that Jeremy's entailment works with Patrick's extra rules
- [15:34:42] bwm
- Jos has rules and a formal proof that Jeremy's entailment works with range and drange with ageA and ageB example - with extra rules
- [15:34:42] bwm
- Patrick suggests we only need one age property but this requires assuming a datatype class includes both value and lexical space members
- [15:34:42] bwm
- Patrick claims with some support from others that this also allows to do without drange
- [15:34:44] bwm
- Having to use two properties is ugly
- [15:34:46] bwm
- We have agreed that the datatype being a union of val and lex spaces is ugly
- [15:34:50] AaronSw
- we can see it online now
- [15:35:08] DanC
- [all can see now]
- [15:35:15] DanC
- chair: this summarizes the discussion earlier today.
- [15:35:36] DanC
- "Jeremy's entailment" refers to...
- [15:35:39]
- * DanC surfs...
- [15:36:00] DanC
- ... PatS's rules [@@pointer pending] were discussed...
- [15:36:01] gk
- Jeremy's entailment was basically that the same info could be entailed from both idioms
- [15:36:34] DanC
- PatS's rules: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0642.html
- [15:37:13] DanC
- Jeremy's entailment: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0635.html
- [15:37:25] gk
- Are you getting this ... bwm just read a key point?
- [15:37:43] DanC
- PatS grins cuz nobody's formally disputed the "you can get away with just one age property if you put both val and lex in the integer class".
- [15:38:27] DanC
- chair finishes discussiong.
- [15:39:12] DanC
- pathayes: why is it considered "ugly" to use ageA vs. ageB?
- [15:39:25] DanC
- DanC: for "ugly" read "hard to explain to users"
- [15:40:09] DanC
- pathayes: seems easy to explain to me; one property talks about numbers, the other about numerals.
- [15:40:57] gk
- DC folks get confused about properties -- sometimes simple values, sometimes have other values
- [15:41:27] gk
- pathayes: can define an RDFS superproperty that covers the two (literal+value) cases
- [15:42:25] DanC
- pathayes: has anybody looked at my latest? I think it handles Jeremy's case without getting into unions.
- [15:42:36] DanC
- JosD: yes, I agree.
- [15:42:46] DanC
- Jeremy: haven't read the latest yet...
- [15:43:07] DanC
- ... this morning, I went from "you can't make my entaliment work" to...
- [15:43:20] DanC
- ... "hmm... maybe you can"
- [15:44:02] DanC
- chair: the latest proposal wasn't the basis of the discussion today, unfortunately./
- [15:44:30] DanC
- chair: perhaps folks could read it[pointer?] tonight.
- [15:44:36] JosD
- read http://www.coginst.uwf.edu/users/phayes/simpledatatype23-02-2002.html
- [15:44:47] DanC
- chair: about this union idea, pat, you've seen problems?
- [15:46:01] DanC
- pathayes: if the datatypes, as classes, have the union of the [scribe in discussion]
- [15:46:15] DanC
- ppl will want to refer to the value space of the datatypes.
- [15:47:14] DanC
- chair: if someone wanted to say "the value space is the value space of integer", there's now way to do that.
- [15:47:21] DanC
- ... is that the problem?
- [15:48:14]
- * DanC reminds all we've got a half-duples connection
- [15:48:56]
- * DanC doesn't hear anything worth scribing; just people re-re-re-re-stating their preference
- [15:50:42] DanC
- pathayes: we have all those inheritance problems...
- [15:51:08] DanC
- PatS: I'm aware of those problems; I pointed them out; I think those problems are lower costs than these.
- [15:51:30] gk
- I thought the real concern here was ageA/ageB rather than range/drange
- [15:53:15] DanC
- points of order are raised...
- [15:53:24] DanC
- ... re difficulty of phone connection.
- [15:53:25] DanC
- 2nded
- [15:53:50] DanC
- chair, pls re-iterate inheritance problem in email, path?
- [15:54:12] DanC
- RESOLVED: all to read pat's proposal:
- [15:54:32] DanC
- http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0643.html
- [15:54:38] DanC
- http://www.coginst.uwf.edu/users/phayes/simpledatatype23-02-2002.html
- [15:54:43] DanC
- for tomorrow.
- [15:54:53] DanC
- moved to adjourn telcon 'till tomorrow,
- [15:54:57] DanC
- in hopes of better facilities.
- [15:55:03] DanC
- with thanks from the chair, telcon ADJOURNED>
- [15:55:07] DanC
- s/>/./
- [15:55:59] pathayes
- pathayes has left #rdfcore
- [15:57:44] pathayes
- pathayes has joined #rdfcore
- [15:58:58] pathayes
- pathayes has left #rdfcore
- [16:00:25] gk
- discussio nof the inheritance problem
- [16:00:31] gk
- intdec range decimaint
- [16:00:34] gk
- inoct range octalint
- [16:00:51] gk
- inoct subtype indec
- [16:01:17] gk
- allows us to conclude that the range of inoct is also range decimaint
- [16:01:18] DanC
- so it seems the problem is in claiming "Octint rdfs:subClassOf DecimalInt".
- [16:01:42] gk
- q whetere the example is ok
- [16:02:14] gk
- as long as we use range, we have this problem and we have to explain it to users in order to avoid this
- [16:02:20] gk
- with drange, this isn't so muich of a problem
- [16:02:35] gk
- bwm: we have a choice
- [16:03:18] DanC
- DanC: is that as bad as the inheritance problems get?
- [16:03:22] DanC
- [several]: yes.
- [16:03:50] gk
- earlier, noted this doesn't require/isn't caused by proposed union (val, lex)
- [16:04:46] DanC
- EricM: how about the case of W3C dates vs. ISO dates? does this problem show up there?
- [16:04:56] DanC
- [several]: doesn't seem like it.
- [16:06:36] DanC
- Jeremy: I apprecaite the aesthetic argument; clarity comes from using different symbols for different things. It seems that the "range" and "Datarange" concepts are different, so I can see the desire to use different symbols, but I can also see the cost to users. [not sure I got that right]
- [16:07:28]
- * DanC wonders if the chair-projector-owner would give a bit more screen space to this log
- [16:07:28] gk
- If people behave logically, there's no problem
- [16:08:00] DanC
- DanC: or, to put it another way: this might be the best formalism of what people do, in practice, when not constrained.
- [16:09:23] gk
- DanC: I disagree... looking at UAProf->CC/PP, I think that what people do when not constrained is don't specify how datatype mappings work. There's a right old mish-mash of property value formats used there.
- [16:11:00] bwm
- bwm:...
- [16:11:20] bwm
- inheritance problem is?
- [16:11:24] bwm
- fragility
- [16:11:37] bwm
- looks fine but a potential minefield (gk)
- [16:12:16] bwm
- em: inheritance powerful but datatypes tend to be a seaprate thing that thee users don't see, but are reused
- [16:12:40] bwm
- .. and the need to extend them seems less important than the need for new classes and properties
- [16:13:35] bwm
- ... maybe have to sometimes have to pragmatically say that magic happens;datatypes might be one of this
- [16:13:44] bwm
- ... keep it simple; keep it consitently simple
- [16:13:56] bwm
- ... or, also do it wrong like others ahve done it wrong
- [16:14:27] bwm
- jjc: ugly (math) versus ugly (to users). The fromer means fragile honed by experience but might break later
- [16:14:34] bwm
- ... a different sort of risk
- [16:14:51] bwm
- gk: we are trying to overload range with too much - subclassing and datatyping
- [16:15:13] bwm
- ... less likely to get caught in the future if/when datatyping and classes need to go inslightly different directions
- [16:15:41] bwm
- ps: not overloaded, but not as preciese or explcit as could be
- [16:16:09] bwm
- .. range isn't overloaded, more not giving the precision of internal structure of datatypes
- [16:16:57] bwm
- miked: in daml+oil, we addressed some of this by having separate datatype properties and objet properties - seemed necessary, see this as some variation. Found a lot of cases where I foundthis distinction an issue (problem)
- [16:17:25] bwm
- em: we can do this by asking users for their feedback ont his diff. What did they say? Did they get it?
- [16:17:32] bwm
- miked: accepted it pretty well
- [16:17:57] bwm
- ... sometimes had to backup to use Property
- [16:18:12] bwm
- ... nice to not have to make this decision
- [16:18:49] bwm
- em: museum community want one type of property, have far richer semantics
- [16:19:23] bwm
- ... the modellers also want few choices to model
- [16:19:37] bwm
- jjc: heard miked say having 2 types of prop was tricky?
- [16:20:08] bwm
- ... think this is an argument for having drange
- [16:20:28] bwm
- gk: wandered off drange/range into two properties agea/ageb
- [16:20:59] gk
- DaveB: we could sell range/drange much more easily than two names for each property.
- [16:21:09] bwm
- jjc: drANGE/range only effects schema authors -e xpertes, but two properties affects all end uers
- [16:21:25] bwm
- JosD: we aren't proposing two properties
- [16:21:42] bwm
- em: what is the vlaue of drange?
- [16:21:48] gk
- jjc: previously, suggested that range/drange may be anough to avoid needing two names for each property. (Didn't catch why.. did I misunderstand?)
- [16:22:10] bwm
- em: in schemawg datatypes dropped..
- [16:22:26] bwm
- ... now returning
- [16:22:51] bwm
- ps: range only versus datatypes as unions -
- [16:22:59] bwm
- .. then all rdffs range semantics remain the same
- [16:23:07] bwm
- ... bt ca't say only value or only lex space
- [16:23:18] bwm
- ... however with drange, can say what a datatypin class
- [16:23:39] bwm
- ... from rdfd:drange it says
- [16:23:50] danbri_
- historical note: they were dropped by RDF Schema WG so we could wait to see what XML schema datatyping looked like
- [16:23:56] bwm
- ... the object of a property with a drange is either a union of a daatvalue or a lexical space
- [16:24:02] bwm
- danbri_: yeah
- [16:24:06] bwm
- (dajobe btw)
- [16:24:13] bwm
- bwm is now known as dajobe-scribe
- [16:24:20] dajobe-scribe
- ... dual view of datatype
- [16:24:47] dajobe-scribe
- jjc: if using drange then opposed using uris for datatype uris confusing if sometimes users have to use range, sometimes drange
- [16:24:54] dajobe-scribe
- ... either range OR drange, not both
- [16:24:59] dajobe-scribe
- .. for datatyping
- [16:25:23] dajobe-scribe
- em: large confusion on users from datatyping, diff from a class, and how to model a datatype / class ..
- [16:25:30] dajobe-scribe
- ... e.g. dewey, basically a float
- [16:25:38] dajobe-scribe
- ... classes and subclasses with relationships
- [16:25:45] dajobe-scribe
- ... but what about controleld vocab?
- [16:25:55] dajobe-scribe
- ... (mon, tue, ... fri) a dataype or a set o clases?
- [16:26:04] dajobe-scribe
- ... no pref, just there are oddities here
- [16:26:24] dajobe-scribe
- ... there are a default set of datatypes that people like and they get confused in the diff/similaritiies#
- [16:26:32] dajobe-scribe
- ... we have to explain clearly the fundamental differences
- [16:26:38] dajobe-scribe
- ... degrees of ugliness again
- [16:27:00] dajobe-scribe
- ... If they are different, we have to be clera when to use which symbol.
- [16:27:25] dajobe-scribe
- ps: one one hand have a min vocab ...
- [16:27:44] dajobe-scribe
- ... they won't understand range/drange
- [16:27:56] dajobe-scribe
- (em: you agreein with me?
- [16:27:58] dajobe-scribe
- ... yes
- [16:28:29] dajobe-scribe
- ... if we give this union vocab, is simple enough. For others, additional machinery can be given for more powerful users
- [16:28:39] dajobe-scribe
- bwm: martyn - when do I use range, when drange?
- [16:28:45] DanC
- DanC has quit
- [16:29:05] dajobe-scribe
- bwm asks around the table
- [16:29:54] dajobe-scribe
- JosD: like daml datatype property; when you want to use xsd: things
- [16:30:01] dajobe-scribe
- ... no free lunch
- [16:30:22] dajobe-scribe
- em: diff is when you do literals or resources for the values
- [16:30:32] dajobe-scribe
- ... for age as an aexample
- [16:30:40] dajobe-scribe
- ... we have two properties
- [16:31:11] dajobe-scribe
- jjc: drange when want to be a literal; range when want to it be an object
- [16:31:29] dajobe-scribe
- ps: in merging graphs of these two styles, get a type clash
- [16:31:41] dajobe-scribe
- ... or always have to say drange
- [16:31:53] dajobe-scribe
- ... get a range conflict
- [16:34:18] dajobe-scribe
- jjc writes an example
- [16:34:43] dajobe-scribe
- asked dajobe-scribe: no answer
- [16:36:07] dajobe-scribe
- bwm: wrap up
- [16:36:43] dajobe-scribe
- ... need to do:
- [16:36:56] dajobe-scribe
- ... 2) where are we gonna go tonight?
- [16:37:11] dajobe-scribe
- ... 1) tomorows agenda
- [16:38:40] dajobe-scribe
- review of 26th agenda
- [16:38:58] dajobe-scribe
- mk: from webont perspsective xml:lang and cardinality constraints?
- [16:39:04] dajobe-scribe
- s/mk/mdean/
- [16:39:09] dajobe-scribe
- jjc: isn't that a webont q?
- [16:39:24] dajobe-scribe
- em: want to talk to you about this
- [16:40:24] dajobe-scribe
- em: vote for xml schema relationships 10 min
- [16:45:06] dajobe-scribe
- agenda review
- [16:45:22] dajobe-scribe
- bwm: try to fit more daatyping in
- [16:46:01] dajobe-scribe
- bwm: regret not bein able to go over pat's document during the meeting
- [16:46:06] dajobe-scribe
- general approval
- [16:46:14] dajobe-scribe
- bwm: read it tonight
- [16:47:02] dajobe-scribe
- CLOSE
- [16:47:06] dajobe-scribe
- dajobe-scribe is now known as bwm
- [16:49:36] libby
- libby has quit
- [17:08:06] bwm
- bwm has quit
- [17:14:53] gk
- gk has left #rdfcore
- [17:14:53] JosD
- JosD has quit
- [17:28:57] mdean_
- mdean_ has quit
- [17:57:58] danbri_
- danbri_ has left #rdfcore
Provided by Dave Beckett,
Institute for Learning and Research Technology, University of Bristol