- From: Dave Beckett <dave.beckett@bristol.ac.uk>
- Date: Wed, 31 Jul 2002 10:09:55 +0100
- To: RDF Core <w3c-rdfcore-wg@w3.org>
- CC: Patrick Stickler <patrick.stickler@nokia.com>
>>>Patrick Stickler said: > It's too bad that the list we produced was not reflected in the official > minutes from the f2f. Perhaps it is somewhere in the IRC logs. Sorry, that's my fault - they were recorded when the IRC logging was down and although I recovered them, I hadn't restored them in-situ. I attach the logs below which include the following list of datatyping things (lightly editing for presentation): 1. Simple inline idiom: <Jenny> <age> "10" . 2. Connolly entailment: <john> <age> "10" . <jenny> <age> "10" . ==> <john> <age> _:x . <jenny> <age> _:x . 3. Explicit representation of a value: <Jenny> <age> _:b . _b <xsd:int> "10" . 4. Cannes entailment: <age> <drange> <xsd:int> . <Jenny> <age> "10" . <john> <age> _:b . _:b <xsd:int> "10" . ==> <Jenny> <age> _:b . 5. Only one age property required. 6. One name for datatype 7. Semantic tidiness 8. Incompatibile datatypes 9. Compatibility with existing practice 10. Query template 11. Minimal machinery 12. Prefer no syntax change Dave --- These are the missing F2F IRC logs for 2002-06-18 The last item it recorded before the bot left was: http://ilrt.org/discovery/chatlogs/rdfcore/2002-06-18.html#T08-32-50 <dajobe-la> added Graph Syntax The first item it recorded after returning: http://ilrt.org/discovery/chatlogs/rdfcore/2002-06-18.html#T09-42-21 The missing items: 08:32:39 <dajobe-la> added Graph Syntax 08:34:02 <dajobe-la> item rdf schema, just done 08:34:08 <dajobe-la> action recorded above 08:34:24 <dajobe-la> item datatypes 08:34:54 --> jjc (~jjc@hplb.hpl.hp.com) has joined #rdfcore 08:35:06 <jjc> nick jjc-scribe 08:35:19 --- jjc is now known as jjc-scrib 08:35:36 <jjc-scrib> Datatypes === Session 6 08:35:52 <jjc-scrib> bwm: let's go over rationale for stake-in-ground 08:36:12 <jjc-scrib> bwm: let's make trade-offs explicit 08:36:34 <jjc-scrib> bwm: we could make different choices if strong agreement 08:36:40 <JosD> * JosD waves to em 08:36:54 <jjc-scrib> bwm: if we rathole then back to stake-in-the-ground or give up 08:36:58 <-- logger_1 has quit (Ping timeout) 08:37:31 <jjc-scrib> bwm continues trying to identify tradeoffs .... 08:37:41 <jjc-scrib> Requirements: 08:37:57 <jjc-scrib> 1. simple inline idiom: <Jenny> <age> "10" . 08:38:02 <dajobe-la> (note logger may not be here, but I'll record the logs publically later) 08:38:58 <jjc-scrib> 2. Connolly entailment: <john> <age> "10"; <jenny> <age> "10" ==> <john> <age> _:x; <jenny> <age> _:x . 08:39:36 --> danbri has joined #rdfcore 08:39:54 <jjc-scrib> 3. Explicit representation of a value: <Jenny> <age> _:b; _b <xsd:int> "10" . 08:42:05 <jjc-scrib> 4. Cannes entailment: <age> <drange> <xsd:int>; <Jenny> <age> "10"; <john> <age> _:b; _:b <xsd:int> "10". ==> <Jenny> <age> _:b . 08:42:23 <jjc-scrib> 5. Only one age property required. 08:42:45 <jjc-scrib> 6. One name for datatype 08:44:34 <jjc-scrib> jos: we could use two names ala bwm a few weeks ago 08:44:45 <jjc-scrib> jjc: what is the uri for xsd:int? 08:45:35 <jjc-scrib> 7. tidiness (both syntactic and semantic) 08:49:30 <-- bwm has quit (Ping timeout) 08:49:36 <jjc-scrib> discussion of syntactic vs semantic tidiness 08:50:41 <dajobe-la> (roll call?) 08:51:01 <jjc-scrib> md: "graph tidiness" not syntactic tidiness 08:51:24 <jjc-scrib> 7. graph tidiness 08:51:29 <jjc-scrib> 8. semantic tidiness 08:52:49 <jjc-scrib> graph tidiness deleted - since the advocates want semantic tidiness 08:53:15 <jjc-scrib> 7. semantic tidiness 08:53:24 <jjc-scrib> 8. incompatibile datatypes 08:53:41 <jjc-scrib> 9. compatibility with existing practice 08:54:42 <jjc-scrib> 10. query template 08:56:11 <jjc-scrib> 11. minimal machinery 08:57:52 <jjc-scrib> 12. Prefer no syntax change 09:03:11 <jjc-scrib> discussion of 2. 09:04:09 <jjc-scrib> path, guha: we have a solution which gives 2 but not 7 (range depends on property) 09:04:43 <jjc-scrib> bwm: how about property rdf:object (from reification vocab) 09:05:27 <jjc-scrib> bwm: currently no constraint that there is an implicit datatype range 09:06:09 <jjc-scrib> guha: connolly entailement is necessary for consistency 09:06:53 <jjc-scrib> connolly entailment is is the interpretation of the triple a fucntion of only the predicate and object 09:09:04 --> bwm has joined #rdfcore 09:09:41 <jjc-scrib> guha: what are the arguments to the denotation of the string 09:09:55 <jjc-scrib> jjc: jenny child, john baby, years vs months 09:10:22 <jjc-scrib> patrick displays document 09:10:35 <jjc-scrib> without URL 09:12:20 <jjc-scrib> guha: consider function df("FFF", s1, NULL) "FFF" is lexial value, s1 is prop, NULL is datatype range 09:13:05 <jjc-scrib> we can work on the desiderata in terms of the arguments to the df function 09:13:59 <jjc-scrib> guha: df is deterministic 09:15:13 <jjc-scrib> mike: s1 is subj or propety. guha: property 09:15:38 <jjc-scrib> sergey: as long as df has more arguments than merely "FFF" then we are in the semantictally untidy 09:16:40 <jjc-scrib> guha: codes for cities from db usage 09:17:23 <jjc-scrib> jos: issue to do with non-monotocity? 09:17:34 <jjc-scrib> path: no NULL is an existential quantifier 09:17:46 <jjc-scrib> guha: agree with pat 09:18:34 <jjc-scrib> jjc: null is quantification over unknown set of datatypes 09:18:49 <jjc-scrib> guha: NULL stands for a datatype but we don't know which 09:20:17 <jjc-scrib> frank: df takes the literal in context and decontextualises it 09:20:27 <jjc-scrib> ^frank^guha 09:24:48 --> bwm_ has joined #rdfcore 09:24:49 <-- bwm has quit (Connection reset by peer) 09:24:59 <jjc-scrib> discussion about how to discuss 09:26:44 <jjc-scrib> sergey: tidiness is crucial issue 09:31:09 <jjc-scrib> gk: tidy case L2V is a fucntion of string 09:33:08 <jjc-scrib> sergey: makes presentation 09:33:30 <jjc-scrib> datatypes are strictly on top of RDF schema 09:34:26 <jjc-scrib> sergey: simplify xsd 4-way dts to a 2-way set of individuals and facets 09:35:39 <jjc-scrib> int example 09:39:02 <jjc-scrib> sergey the facets of integers can be mapped onto facets of lexical tokens that correspond to integers 09:39:16 <jjc-scrib> gk: may i test my understanding 09:40:01 <jjc-scrib> gk: are datatypes in domain of discourse? 09:41:05 <jjc-scrib> patrick: RDF does not care about facets 09:41:24 <jjc-scrib> sergey: facets determine the semantics of a datatype 09:41:39 --> logger_1 has joined #rdfcore 09:42:10 <jjc-scrib> path: are we talikng about the value spac or the lexical space
Received on Wednesday, 31 July 2002 05:12:28 UTC