W3C WebOnt Working Group IRC logs for 2002-01-15

These are the automatically generated logs from the W3C WebOnt Working Group IRC chat

Dave Beckett


You are here: Logs Home / 2002-01-15

Latest logs

[00:39:53] ora
ora has left #webont
[02:16:59] sandro
sandro has quit
[02:25:51] sandro
sandro has joined #webont
[11:54:05] logger_1
logger_1 has joined #webont
[11:54:05]
Topic now WebOnt ftf, NJ http://www.w3.org/2001/sw/WebOnt/ftf1.html
[11:54:05]
Users on #webont: logger_1 sandro em @logger
[11:54:06] logger
logger has quit
[11:58:10] logger_3
logger_3 has joined #webont
[11:58:10]
Topic now WebOnt ftf, NJ http://www.w3.org/2001/sw/WebOnt/ftf1.html
[11:58:10]
Users on #webont: logger_3 logger_1 em sandro
[11:58:11] logger_1
logger_1 has quit
[12:01:04] logger_2
logger_2 has joined #webont
[12:01:04]
Topic now WebOnt ftf, NJ http://www.w3.org/2001/sw/WebOnt/ftf1.html
[12:01:04]
Users on #webont: logger_2 em sandro
[12:11:16] em
em has quit
[12:11:58] em
em has joined #webont
[13:52:19] DanC
DanC has joined #webont
[13:59:56] nmg
nmg has joined #webont
[14:02:25] jdale
jdale has joined #webont
[14:09:07] DanC
===== day 2 gets started
[14:09:13] DanC
... ideas on next ftf
[14:09:25] DanC
one option: Eu late april.
[14:09:31] DanC
another option: Hawaii in May
[14:10:23] DanC
Feb tech plenary is too soon for a ftf WG meeting, but you're welcome to come to the RDF/Semantic Web IG meeting there.
[14:11:20] nmg
EU meeting would be colocated with KR2002 - http://www.kr.org/kr/kr02/
[14:12:06] nmg
(which itself is colocated with DL2002 - http://www.cs.man.ac.uk/dl2002/)
[14:13:54] jdale
Fujitsu could still offer to host it in northern california
[14:15:13] nmg
Mike Dean's slides:
[14:15:15] nmg
http://www.daml.org/2002/01/experiences/
[14:17:05] DanC
A:why does reflexe cause complexity problem?
[14:17:12] DanC
er.. Q: why...
[14:17:25] DanC
A: because it breaks the tree... [?]
[14:20:20] DanC
===== DAML Experience reports, Mike Dean
[14:21:43] TimFinin
TimFinin has joined #webont
[14:21:48] DanC
Mike Dean presents from "DAML+OIL Issues and Experiences" http://www.daml.org/2002/01/experiences/
[14:22:33] DanC
Q: "frame orientation"... what does that mean?
[14:23:04] DanC
A:chunking is part of it... in DAML+OIL, there's no way to say "this chunk is my definition of this class"
[14:23:16] JosD
JosD has joined #webont
[14:23:32] DanC
A:also, unnamed classes and class expressions are a different way of thinking
[14:24:06] libby
libby has joined #webont
[14:24:33] DanC
A:to some extent, we've built tools that make DAML+OIL look like a frame system, but there are some language issues that break the illusion
[14:28:05] DanC
Q:don't frame systems usually include defaults and such non-monotonic stuff?
[14:29:19] DanC
A:so let's apply the 80/20 rule: take the stuff from frame systems that's really useful and easy...
[14:31:16] DanC
DebM: I've done DL research for a long time and I now have all these frame editors... I just see them as interfaces to DL systems where it's inconvenient to make defined classes.
[14:31:41] DanC
[...scribe isn't sure where we're headed and what to capture...]
[14:35:10] DanC
[a few]: It seems like there should be more stuff in the language to capture higher-level idioms corresponding to frames etc.
[14:37:58] DanC
[... UML syntax/idioms are also part of the discussion...]
[14:42:15] DanC
Chair: there are at least two different user communities, and we should be aware of the needs (including documentation, ...) of the various communities
[14:42:37] jah-wowg
jah-wowg has joined #webont
[14:42:38] DeborahMcGuinness
DeborahMcGuinness has joined #webont
[14:43:12] DanC
IanH: I heard a consensus around extra-logical features to support grouping of properties with classes and such.
[14:43:36] DeborahMcGuinness
i just got on sorry - one note for the scribe is to do meta tagging for properties associated with term definitions
[14:44:19] laurentO
laurentO has joined #webont
[14:46:30] DanC
folks are welcome to discuss in a break, send proposals, etc.
[14:53:17] DanC
JimH: let's be very careful about saying "our language can express that; we don't need to add it"... even though UnambiguousProperty is expressible in terms of cardinality constraints, that wouldn't make UnambiguousProperty visible to users
[14:58:59] DanC
as MikeD gets to rules...
[15:00:31] DanC
Chair clarifies our charter w.r.t. rules: it's recognized that rules are needed in the Semantic Web, but ontologies are speparable, and there's more consensus on the technical design at the ontology level
[15:01:29] DanC
there are other fora (www-rdf-rules, RDF IG ftf in Feb, ...) to discuss rules.
[15:30:26] libby
libby has quit
[15:35:53] jah-wowg
logger, pointer?
[15:36:05] jah-wowg
logger_2, pointer?
[15:36:05] jah-wowg
See http://ilrt.org/discovery/chatlogs/webont/2002-01-15#T15-36-05
[15:37:22] jdale
jdale has quit
[15:48:26] nmg
nmg has quit
[15:48:27] JosD
JosD has quit
[15:48:29] TimFinin
TimFinin has quit
[15:48:54] DeborahMcGuinness
DeborahMcGuinness has quit
[16:38:56] laurentO
laurentO has quit
[17:28:50] ora
ora has joined #webont
[17:40:47] jdale
jdale has joined #webont
[17:47:13] jah-wowg
jah-wowg has quit
[18:12:05] DanC
===== reconvene after lunch
[18:12:22] DanC
new agenda (ftf-2002/slide1-2.html on hendler's machine)
[18:12:45] DanC
hot topics: Decidability, RDF compatibility [aka layering]
[18:15:29] DanC
=== next meeting
[18:15:40] DanC
option: 25/26 Mar, Europe
[18:15:49] DanC
option: 1-2 April US East
[18:15:57] DanC
option: 8-9 April. US East
[18:16:39] DanC
most likely seems to be 25/26 Mar in Amsterdam
[18:17:27] TimFinin
TimFinin has joined #webont
[18:17:51] DanC
===== Requirements editors
[18:17:58] DanC
Hefflin have volunteered to co-edit
[18:18:07] DanC
Jonathan and Rafael volunteer too.
[18:18:59] DanC
[JimH starts editing REQUIREMENTS list in a text editor]
[18:19:17] DanC
-- Collection management
[18:19:57] DanC
archtypical use cases: (a) collection of web pages in a web site. (b) homogeneous collection: art, etc. (c) documents around a physical artifact; e.g. an engineering design for [an aircraft?]
[18:21:37] nmg
nmg has joined #webont
[18:23:17] DanC
not a requirement: more complex constraints than range constraints on datatypes
[18:25:15] DanC
[4 requirements on the board]
[18:25:59] DanC
MikeD on "content interoperability"
[18:26:12] DanC
top use cases: travel planning/[agency?]
[18:26:17] DanC
^1
[18:26:21] DanC
2. ontoweb portal
[18:30:28] DeborahMcGuinness
DeborahMcGuinness has joined #webont
[18:31:59] DanC
-- Decker, web services
[18:32:13] DanC
use case: ubiquitous computing: small devices, mutual authentication, ...
[18:41:38] sandro
sandro has quit
[18:42:31] sandro
sandro has joined #webont
[19:00:20] nmg
nmg has quit
[19:00:29] TimFinin
TimFinin has quit
[19:00:33] DeborahMcGuinness
DeborahMcGuinness has quit
[19:23:41] ora
ora has quit
[19:24:00] ora
ora has joined #webont
[19:36:33] DanC
ACTION Hefflin, Nick G., Rafael V.: to draft a requirements document by end of jan.
[19:36:37] DanC
oops...
[19:36:47] DanC
ACTION Hefflin, Jonathan D., Rafael V.: to draft a requirements document by end of jan.
[19:40:31] libby
libby has joined #webont
[19:40:35] DanC
NOTE: official publication rules: http://www.w3.org/Guide/pubrules
[19:41:48] DanC
ACTION JimH: send table of requirements from discussion before that break
[19:41:59] DanC
[resume from break]
[19:43:04] TimFinin
TimFinin has joined #webont
[19:47:11] TimFinin
finin will scribe
[19:47:32] TimFinin
discussion on decidability as a requirement
[19:47:57] TimFinin
how important is it?
[19:49:08] TimFinin
what's the cost of decidabiltyu? danc: costs include ruling some features in and some features out
[19:49:57] TimFinin
fvh: what are the reasoning tasks that ppl want to do? it is for thoise tasks that we have to ask about decidability
[19:50:29] TimFinin
jimh: for daml+oil, decidability was a critereon when evaluating a proposed feature.
[19:50:52] TimFinin
jimh: we are now addressing a large community.
[19:51:25] TimFinin
jimh: some subcommunities need a richer representation and don't need some reasoning features, like subsumption
[19:52:10] TimFinin
danc: can anyone speak to the reasoning tasks that have to be decidable?
[19:54:48] TimFinin
jimh and ih: were really talking about computability
[19:55:57] TimFinin
ih: subsumption checking is needed by ppl who are building large ontologies with multiple authors.
[19:56:48] TimFinin
lh: for example, i'm working with users who are trying to build ontologies for genomics
[19:57:09] TimFinin
danh: is it unaceptable if the reasoning is incomplete?
[19:57:14] TimFinin
ih: it depends
[19:58:20] TimFinin
jeramy: suggests that we might have decisbility as a requirement for OWL 1.0. we can relax it later.
[19:59:11] TimFinin
gus: I have one use case involving matching descriptions of stolen art objects
[19:59:37] TimFinin
the use case could go either way -- using subsumption or something less.
[20:02:05] TimFinin
jimh: the NCI uses two tools to build ontologies -- a DL like system and a more proceedural system. they seem to need both.
[20:04:32] TimFinin
tf: worse is better
[20:05:08] TimFinin
fvh: you can still write useful tools to do things even if the language is undecidable.
[20:05:50] TimFinin
jimh: I'm hearing efficiency as a goal, but not a strict reaquirement.
[20:07:23] TimFinin
pfps: those of us who have to build a reasoner to handle the language know how to do some things but not others.
[20:09:06] TimFinin
pfps: two examples of communities who want subsumption: (1) medical informatics people with lots of data (2) trael services.
[20:13:37] TimFinin
TimFinin has quit
[20:13:37] ora
ora has quit
[20:14:57] ora
ora has joined #webont
[20:15:56] DanC
tim finin is now using danc's machinr
[20:16:53] DanC
pfps: we need an entailment reasoner.
[20:17:58] DanC
jimh is looking for the right descriptive term for the reasoner we want.
[20:18:09] DanC
ih suggests "class consistancy reasoner"
[20:18:47] TimFinin
TimFinin has joined #webont
[20:19:17] DanC
ian and frank are delegated the task to come up with the descriptive adj for the reasoner.
[20:19:44] DanC
the next period will cover the discussion of rdf compatibility.
[20:20:27] DanC
jimh: pfps described five attributes (babies) all of which together cause problems. which to throw out?
[20:20:56] TimFinin
TimFinin has quit
[20:23:28] DanC
reconstucting peter's presentation from yesterday, there are three possibilities.
[20:23:49] DanC
(1) a syntactic embedding of owl into rdfs
[20:24:48] DanC
(2) owl has syntactic features that go beyond rdfs
[20:28:25] DanC
(3) owl is syntactically like rdfs but has different semantics
[20:35:59] DanC
a fourth possibility is peter's owl', in which owl' is a semantic restriction of a syntactic restriction of rdfs
[20:36:02] DanC
danc does a staw poll on the three suggestions.
[20:36:12] DanC
option (1) had eight people who liked it and 1 who found it unattractive
[20:36:14] DanC
(2) had two who liked it and 2 who disliked it
[20:36:22] DanC
(3) had one who liked it and 7 who disliked it.
[20:43:04] DanC
discussion of the tbl layer cake. danc says that tbl's vision is for option (3).
[20:43:05] ora
I am wondering whether DAML+OIL is considered like option #2 even though the syntactic incompatibilities are within the standard RDF syntax extension framework
[20:43:16] DanC
jimh: were running out of time and patience.
[20:47:53] DanC
the babies: meta-modelling; negation; entailment; extra syntax
[20:53:33] libby
libby has quit
[20:56:49] DanC
[20:58:07] DanC
now lets discuss the use cases that make each of the possibilities bad
[20:59:53] DanC
for (1) owl and rdfs produce overlaping sents of sentences. this means that we may need to know whether a document
[21:00:05] DanC
was intended to be processed via owl or rdfs.
[21:02:33] DanC
how will we go forward on this issue?
[21:03:30] DanC
it's proposed that several people go off and write a document describing the options, what the problems are and what's needed to fix the problems.
[21:04:56] DanC
pfps and dieter are likely candidates
[21:07:46] DanC
mike smith has been volunteered and his weak objections easily overcome by collective pressure.
[21:08:18] DanC
ACTION PeterPS, Dieter, Mike Smith: write up layering issues
[21:08:51] DanC
by end of JAn
[21:09:23] DanC
ziv helmond has been added to the group.
[21:11:32] DanC
jimh is concerned that we don't have a process in place for beginning to write down the specification
[21:12:39] DanC
danc suggests starting ith a tutorial type document. a cookbook. written in prose with lots of examples.
[21:15:59] DanC
jimh suggests we look at how daml+oil reflects the requirements that we've generated in this f2f meeting
[21:16:53] DanC
danc is willing to try to do this.
[21:17:17] DanC
jimh believes that Ian should be in the loop on this and ian concures.
[21:19:31] DanC
ACTION: danc Ian and mike Smith will work on a document which evaluates how well daml+oil meets the owl requirements as identified at this f2f meeting
[21:22:05] DanC
jimh thinks we may be about done.
[21:23:04] DanC
jimh three action items: (1) requirements document (2) owl write up on layering issues and (3) mapping of requirments to daml+oil.
[21:23:37] DanC
immediate action item: plan meetings for next year?
[21:23:54] DanC
ACTION deadline for document #3 is January 31, 2002.
[21:28:51] DanC
we're looking at the calendar for 2002 and talking about when to have upcoming meetings.
[21:29:09] DanC
we're anticipating having a meeting every three months or so.
[21:40:57] DanC
some suggestions for f2f meetings: #2 on March 25-26, #3 just before iswc june 6-7
[21:42:07] DanC
meeting 33 might bein late sept or early october.
[21:44:01] DanC
meetings #2 and #3 look like europe (amsterdam for 2 and sardinia for 3) so meeting four might be in the US in october
[21:50:52] DanC
Now it's suggested that we hold f2f #2 april 8-9 in the us east coast in conjunction with the semantic web related meeting in Atlanta
[21:51:32] DanC
ACTION JimH: send ftf schedule proposal
[21:52:23] DanC
moved to ajourn. seconded.
[21:52:37] DanC
eof
[21:53:43] jdale
jdale has quit
[22:27:22] DanC
DanC has quit
[23:52:19] sandro
sandro has quit
[23:52:19] ora
ora has quit
[23:52:49] sandro
sandro has joined #webont

Provided by Dave Beckett, Institute for Learning and Research Technology, University of Bristol