WWW2003 Conference IRC Chat Logs for 2003-05-24
WWW2003 Conference IRC Chat Logs for 2003-05-24 |
|
This web site provides the chat logs for the WWW2003 Community Hub for attendee-generated notes and coverage of the Twelfth International World Wide Web Conference.
You are here: Logs Home / 2003-05-24
07:05:55 <edd> edd has changed the topic to: WWW2003, logged and blogged at http://www2003.xmlhack.com/ - Budapest, Hungary - May 20-24 - Developers Day
07:06:35 <sandro>http://www.w3.org/2003/05/24-semweb-dd/
07:06:36 <dc_www2003> A: http://www.w3.org/2003/05/24-semweb-dd/ from sandro
07:07:05 <sandro> A:| Updates SW DevDay schedule, with morning changes
07:07:06 <dc_www2003> Titled item A.
07:09:49 <dajobe> hmm, where to start typing on semweb dd ?
07:09:58 <edd> sw - guha presenting on tap
07:10:14 * dajobe thanks edd
07:10:34 <edd> guha - strong feelings on keeping things very very simple (yay!)
07:11:07 <dajobe> sw - as easy to publish graph, as html
07:11:17 <dajobe> sw - aggregation as analog of index.html
07:12:05 <edd> question - writing the data is hard
07:12:17 <edd> question - may be eay to serve, but creating it is the hard problem
07:12:46 <edd> guha - mentions tool for mapping database to rdf
07:13:27 * DanC waves from mobile web session
07:13:42 <edd> guha - appears to be proposing a new HTTP verb GetData as an analog of GET for html?
07:13:53 * edd confused if that's what he's really proposing
07:13:59 * mattb not sure why GET can't be a GET for RDF
07:14:08 <DanC> yes, let's just use GET, please!
07:14:18 <DanC> just do GET /fooKB?query-goes-here
07:14:23 <edd> oh, it's a SOAP method
07:14:24 <mattb> iirc from the tapache site it's a soap call
07:15:01 <dajobe> he's talking at the API (code) level; I think there is an HTTP GET i/f
07:15:06 <DanC> danbri and I started this discussed this with the TAP folks; they said something about their queries not fitting in the p1=v1&p2=v2 syntax; I didn't get around to replying, but there's no reason to constrain yourself to that syntax
07:15:23 <dajobe> let's ask
07:15:52 <AndyS> It is a SOAP call with a small graph to identify the thing you want (data returned decided by server)
07:17:37 <DanC>http://lists.w3.org/Archives/Public/public-rdf-tap/2003Jan/0021.html
07:17:38 <dc_www2003> B: http://lists.w3.org/Archives/Public/public-rdf-tap/2003Jan/0021.html from DanC
07:17:41 <sandro> Guha: we use "regular path expressions" to have nice computational properties
07:17:50 <DanC> B:|TAP's use of SOAP: HTTP GET or PUT?
07:17:50 <dc_www2003> Titled item B.
07:17:53 <sandro> TimBL: I think general graph match is simpler, actually
07:18:00 <DanC> B:Dan Brickley (Tue, Jan 28 2003)
07:18:00 <dc_www2003> Added comment B1.
07:18:19 <sandro> Guha: There is a strong tension between them.
07:18:52 <sandro> Guha: "Semantic Negotiation" to identify nodes via description, cached, ....
07:18:55 <DanC> B:McCool said something about their queries not fitting in the p1=v1&p2=v2 syntax; I didn't get around to replying, but there's no reason to constrain yourself to that syntax
07:18:56 <dc_www2003> Added comment B2.
07:19:25 <sandro> Audience: there are 5 RDF query languages that are restricted to path expressions (as opposed to graph match)
07:19:51 <timbl> I feel it would be simpler if the constrints he puts on teh graph are complicated and arbitrary.
07:20:15 <timbl> s/Audience/Bijan/
07:20:35 * edd doesn't apply the regex in general :)
07:21:25 <sandro> perhaps a good compromise: regular path expressions as a performance-profile for graph match.
07:21:42 <edd> guha introduces the problem of establishing trust
07:22:18 <edd> "it's really useful to have large knowledge bases lying around"
07:24:07 * edd would like to see some sort of semantic web user interface SIG. i think we've invented ourselves a bunch of UI problems we need to understand
07:25:34 <sandro> Guha: we have Self Repairing Scrapers, which see when a page format changes and used the connection between the data on the old form and the data on the new one to see how it changed. [ Wow: brilliant. Nearly patent-worthy idea. ]
07:25:58 <edd> Walter Perry is keen on such stuff.
07:27:29 <sandro>http://tap.stanford.edu
07:27:29 <dc_www2003> C: http://tap.stanford.edu from sandro
07:28:10 <sandro> C:| The work Guha is presenting; Get it, Help us improve it
07:28:10 <dc_www2003> Titled item C.
07:28:28 <sandro> C:| The work Guha is presenting; ("get it and help us improve it")
07:28:29 <dc_www2003> Titled item C.
07:29:05 <edd> "we need a good ontology which has no strings attached"
07:29:15 <edd> wild applause
07:29:28 <edd> emiller "many ontologies'
07:29:52 <yod> that was jim hendler, no?
07:29:58 <AndyS> s/emiller/jhendler/
07:30:08 <edd> sorry. cool.
07:30:27 <DanC> the SUMO ontology has the usual "big huge ontology in the sky" problems, but (a) it's modularized into ~10 pieces, (b) there's a nifty mapping from wordnet, and (c) the owners are license-friendly
07:30:32 <AndyS> Guha - "one is better than zero"
07:31:12 <edd> next up jhendler and mindswap team on the implementation of the sci-am article scenario
07:31:52 <DanC> re "good ontology with no strings attached", I'd like to work in that direction. cf http://esw.w3.org/topic/SeedApplications
07:34:27 * edd notes this scenario wouldn't occur in UK healthcare: you better take what they tell you or *else* :)
07:35:27 <edd> jhendler "using the tools we have availble, how much of this could we do now"
07:37:18 <edd> showing PC and handheld PDA interacting. phone call emulated on PDA, music silenced on PC.
07:37:35 <AndyS> uPnP used
07:37:37 * edd can't quite hear right, can someone add in the acronyms of the technologies
07:38:14 <AndyS> uPnP - 'Universal Plug and Play'
07:38:21 <dajobe> both dxevices are deswcribed with daml-s services
07:38:32 <dajobe> that say they are noise producing devices in a gneneric way
07:38:41 <edd> doctor treatment course described in daml-s process model
07:38:43 <sandro> Bijan: treatment plan is a daml-s process model
07:39:09 <sandro> ...: Every doctor will hire grest hacker receptionists. ( joke )
07:41:27 <AndyS> Ontology-driven form building
07:44:04 <sandro> modeling everything a world-altering web-services
07:44:20 <sandro> also using information-gathering (providing) services
07:45:03 <sandro> simulate-in-advance your world-altering services, at plan time.
07:45:21 <sandro> planner has to know which are which
07:45:57 <sandro> DAML-S doesnt have that bit built in, it's an upper ontology, but there's a place to extend it.
07:46:20 <sandro> TimBL: this is a really important bit, why isn't it built in?
07:46:44 <sandro> Bijan: Because it's not a perfectly crisp distinction. Every service alters the world a little bit.
07:47:47 <sandro> Audience: you're nailed to one ontology
07:48:13 <sandro> Bijan: We're using one, coming from the treatment-plan, for this particular planning task, but that's all.
07:51:27 <sandro> Bijan: The medical planner sugested changing one of Lucy's one personal low-priority calendar events
07:51:37 <sandro> s/one/own
07:53:46 <sandro> Hrm. As a user, I sure hope this is reversable. Direct manipulation would be even nicer. (this is the old anti-agent argument, classically attributed to Ben Schneiderman.)
07:58:34 <sandro>http://www.cs.umd.edu/projects/shop/
07:58:35 <dc_www2003> D: http://www.cs.umd.edu/projects/shop/ from sandro
07:59:05 <sandro> D:|SHOP, a good, general planner used for this SciAm demo
07:59:05 <dc_www2003> Titled item D.
07:59:33 <dajobe> "simplistic things are pretty cool"
07:59:51 <dajobe> bijan shows the service composition demo
08:00:01 <dajobe> Fujitsu Steer
08:00:11 <dajobe> men of services to choose from
08:00:14 <dajobe> menu
08:00:35 <sandro> "men of services". Hm. :)
08:00:37 <dajobe> starts with a service that has no input
08:01:03 <dajobe> a set of contact services - that are self evalutating
08:01:36 <dajobe> wrote a translation service that converts one ont's contact addr to annother
08:01:38 <dajobe> FLA2UMD
08:02:00 <dajobe> adds more matches to the "services that have njo input" start
08:24:53 * DanC wonders when the break is scheduled
08:38:37 <olivier> olivier is now known as yod
08:48:55 <AndyS> There have been a few network problems here so there was an early coffee break. Resuming about now.
08:49:28 <dajobe> dom - w3c
08:49:36 * DanC wonders if the slides are available via http
08:49:37 <dajobe> sedmantic web apps at w3c
08:50:00 <dajobe> going to talk about SW app not because he's trying to promite it or show how cool it is
08:50:07 <dajobe> but because it *worked for me*
08:50:23 <dajobe> [laughter - emiller]
08:50:41 <yod>http://www.w3.org/2003/Talks/www2003-devday-semweb/
08:50:41 <dajobe> slide3
08:50:42 <dc_www2003> E: http://www.w3.org/2003/Talks/www2003-devday-semweb/ from yod
08:51:02 <yod> E:| Semantic Web Applications at W3C
08:51:02 <yod> by Dominique Hazael-Massieux
08:51:03 <dc_www2003> Titled item E.
08:51:11 <yod> E: by Dominique Hazael-Massieux
08:51:12 <dc_www2003> Added comment E1.
08:51:20 <dajobe> slide4
08:51:31 <dajobe> TR=technical reports
08:52:08 <dajobe> big digital library
08:52:25 <dajobe> will try to explain the benefits and the process
08:52:26 <dajobe> slide 5
08:52:39 <dajobe> when a wg wants to publish a tech report, there are publication rules (pub rules)
08:52:57 <dajobe> these impose retrictions on the report that it is easy to extract them with xslt
08:53:00 <dajobe> to give the rdf metadata
08:53:09 <dajobe> we combine this new metadata with a list of TR in RDF
08:53:21 <dajobe> and using the easy merging of rdf and processing rules to updated tr
08:54:09 <dajobe> big part of the project has been to get the TR detail in RDF from the older hand-maintained page
08:54:20 <dajobe> following "real process" link
08:54:24 <dajobe> which has a bit more detail
08:54:52 <dajobe> slide6
08:55:10 <dajobe> I used to be the webmaster for w3c, responsible for pubbing the tech reports
08:55:28 <dajobe> agree with the PoV on slide6
08:55:42 <timbl> dom: very differenmt from the usual webmaster job
08:56:03 <DanC> E:wow... time flies when you're having fun hacking... [http://lists.w3.org/Archives/Member/w3c-semweb-ad/2001Oct/0021.html|original project description for TR automation] (member-only; maybe I'll forward to www-archive) dates from Oct 2001.
08:56:04 <dc_www2003> Added comment E2.
08:56:04 <dajobe> pub rate growing, used to be ok when had small number of docs
08:56:58 <dajobe> new views - switches to /TR/tr-editor
08:57:08 <dajobe> using xslt on the RDF file
08:57:14 <dajobe> [wow]
08:57:19 <dajobe> slide7
08:57:39 <dajobe> imporatnt benefits we get dig lib in a formalized way
08:57:57 <dajobe> very decenterialised, benefits - dist maintaenance by people with diff needs
08:58:05 <dajobe> and data can be merged
08:58:11 <dajobe> dom shows the rec-history
08:58:23 <dajobe> . www.w3.org/2003/05/tr-history/rec-history.png
08:58:30 <dajobe> [typed, not cheched]
08:58:43 <dajobe> timeline of spec production activity
08:58:48 <dajobe> very intergate view of the work ion the w3c
08:59:05 <dajobe> as graphical (SVG) easy to see process stages
08:59:10 <dajobe> back to slide7
08:59:19 <timbl> no, http://www.w3.org/2003/05/tr-history/rec-history.svg
08:59:50 <timbl> (he wouldn't ahve eben able to scale a .png like that!)
08:59:51 <dajobe> I think there is a link in slide7
09:00:11 <dajobe> some peple, want an authroative list of our standards
09:00:16 <dajobe> now in RDF, anyone can use it
09:00:18 <sandro> devday room is blocked on port 80, right?
09:00:24 <edd> anyone got the URI of the RDF TR list?
09:00:25 <dajobe> dom switches to show the rdf source of the tr.rdf file
09:00:27 <edd> can't see it from here
09:00:42 <dajobe> it's 2002/01/tr-automation/tr.rdf
09:00:45 <DanC> hmm... a schema diagram would be handy.
09:00:52 <AndyS> DC+local properties
09:00:57 <dajobe> back to slide7
09:01:00 <dajobe> slide8
09:01:07 <DanC> slide 8: A basis for other works
09:02:19 <dajobe> maintainer of the qa matrix only has to maintain what is for him/her
09:03:07 <dajobe> dom shows the translation overview by lang page
09:04:19 <dajobe>http://www.w3.org/2003/03/Translations/OverviewLang.html
09:04:20 <dc_www2003> F: http://www.w3.org/2003/03/Translations/OverviewLang.html from dajobe
09:04:32 <dajobe> F:|W3C Translations listed by Languages and Technologies
09:04:32 <dc_www2003> Titled item F.
09:04:46 <timbl> [discussion devolves into right-left renedering and unicode for a moment]
09:04:48 <dajobe> slide9 related to
09:05:42 * edd browses the tr.rdf file. It would be interesting to have an ontology for them.
09:06:00 <edd> my use case is so XML.com could pull out the ones of interest to its readers
09:06:25 <dajobe> dom shows the ow3c overview SVG
09:06:31 <DanC> i.e. http://www.w3.org/2003/02/W3COrg.svg
09:06:33 <dajobe> with lots of coloured rectangles
09:06:45 <dajobe> dependenies pop up when you mouse over a box
09:06:46 <libby> ooh
09:07:01 <dajobe> [the curly font is a bit hard to read]
09:07:11 <dajobe> "it's cool too for doing some real work"
09:07:15 <dajobe> its
09:07:24 <dajobe> emiller: "very helpful for coordination"
09:07:30 <dajobe> back to slide9
09:08:04 <dajobe> slide9 all about integration
09:08:08 <dajobe> slide10
09:08:26 <dajobe> one thing I really like is that it's really helpful for integration
09:08:31 <dajobe> all the I18N for free
09:09:09 <dajobe> xslt lets you have great power in manipulating your data, such as to svg, and mapping both ways
09:09:27 <dajobe> dom shows the rec process PNG made from rdfs, indirectly
09:09:36 <dajobe> slide11
09:09:45 <DanC> the same RDF file is used (a) as a schema for stuff like rec:WD and such *and* to build that process diagram.
09:09:45 <Jhendler> I wish we had better tools for displaying these RDF graphs when they get larger -- nodes and links get useless fast - some mix of windows and links would be great
09:10:01 <dajobe> Jhendler: guha had one of those in 1995
09:10:03 <dajobe> :)
09:10:26 <dajobe> slide12 decentralized data man
09:10:33 * DanC was plaing with xcruise yesterday, zooming around the filesystem, thinking about curising around RDF graphs
09:10:38 <dajobe> using these technologies, each person is responsible for their own data
09:10:57 <dajobe> and don't have to coordinate tightly, just need to respect the model
09:11:18 <dajobe> one thing that is important is that many of the data here have been provided by other people, not me
09:11:34 <dajobe> slide13 decentralized data manipulation
09:11:54 * dajobe grins at the power of "log:semantics"
09:12:28 <dajobe> why it works so well for us is that the keys are uris
09:12:31 <dajobe> ...
09:12:44 <DanC> log:semantics = http GET + rdf parsing, pretty much.
09:12:46 <dajobe> documenht() in XSLT and the href in XHTML
09:12:54 <dajobe> slide14 upcoming uses
09:13:03 <dajobe> first steps, we have many more ideas
09:13:52 <dajobe> slide15
09:14:15 <dajobe> -- end
09:14:28 * DanC noodles on RDF data from our log files: we already generate a "top 20 invalid pages" and "top 100 pages"; we could maybe export those in RDF, along with info about which links are hot etc.
09:14:51 <dajobe> q&a - can you give an example of what would have not possible if using XML just rather than rDF?
09:15:00 <dajobe> a - schema
09:15:20 <dajobe> relating to things with real properties and uris
09:15:56 <dajobe> modelling. possiblity to work with inference and rules
09:16:01 <dajobe> everything is integrate
09:16:07 <dajobe> the more your formalize, the more integration you can get
09:16:14 <dajobe> emiller: data reuse a huge hit for us
09:16:28 <dajobe> DanC - merging two clouds of data
09:16:36 <DanC> er... files.
09:16:45 <dajobe> next talk
09:17:01 <dajobe> Alberto Reggiori, @semantics
09:17:08 <dajobe> two RdQL application cases
09:17:44 <dajobe> our hypothesis it's possible to build simple semantic web apps today uing RDF, RDQL
09:17:49 <dajobe> they can start simple and grow or change later
09:18:04 <DanC> BLURB: two RdQL application cases
09:18:05 <dc_www2003> G: two RdQL application cases from DanC
09:18:19 <DanC> G:presented by Alberto Reggiori, @semantics
09:18:19 <dc_www2003> Added comment G1.
09:18:23 <DanC> logger_www2003, pointer?
09:18:23 <DanC> See http://ilrt.org/discovery/chatlogs/www2003/2003-05-24#T09-18-23
09:18:34 <dajobe> G:from [http://www.asemantics.com/|@semantics]
09:18:34 <dc_www2003> Added comment G2.
09:18:41 <DanC> G:more [http://ilrt.org/discovery/chatlogs/www2003/2003-05-24#T09-18-23|notes]
09:18:41 <dc_www2003> Added comment G3.
09:18:46 <dajobe> from perl, java apache
09:18:47 <dajobe> two use case
09:18:51 <dajobe> 1) image showcase (ISC)
09:18:57 <dajobe> developed for the ESA/ESRIN
09:19:11 <dajobe> several web collections of text and images that are annotated in several ways
09:19:18 * em-lap grins at the term 'rdfyed'
09:19:32 <dajobe> rdql used to present them
09:19:38 <dajobe> case 2) last minute news (LMN)
09:19:45 <dajobe> protoptype of real-time news clipping servicde
09:19:48 <DanC> "RDFyed"???
09:19:49 <dajobe> using rss, but more rdf than rss
09:19:58 <dajobe> several news formats formatted in m ways scraped in raeal time
09:19:59 <dajobe> also RDFyed
09:20:05 <dajobe> rdql used to buiold display
09:20:09 <dajobe> too fast for me to get urls
09:20:17 <dajobe> but they are off demo.asemantics.com
09:20:30 <dajobe> alberto shows the isc demo
09:20:40 <dajobe> I see demo.asementics.com/biz/isc/
09:20:45 <dajobe> eek, fix that url
09:21:09 <dajobe> hey, there's an [RDF] icon
09:21:14 <dajobe> shows orginal page on esa.it
09:21:33 <dajobe> searches for "france"
09:21:54 <dajobe> and the map changes to show the map with overlayed areas that c over france
09:22:15 <dajobe> nice thing is that everything is there - images are there, where are the productsl and the list of items on thje left
09:22:18 <dajobe> similar for ESA
09:22:26 <dajobe> this is /biz/isc/blue.pl
09:22:34 <dajobe> alberto shows the RDF
09:22:49 <dajobe> description of the original HTML URL
09:23:03 <dajobe> lots of dublin core terms, extent, format
09:23:09 <dajobe> back to the blue.pl interface
09:23:19 <dajobe> searches for italy dublin
09:23:32 <dajobe> third demo
09:23:36 <dajobe> this is /biz/lmn/
09:24:19 <dajobe> G:demo 1 [http://demo.asemantics.com/biz/isc/|ISC]
09:24:19 <dc_www2003> Added comment G4.
09:24:36 <dajobe> alberto shows the "funny articles"
09:24:47 <dajobe> searches for prodi
09:25:24 <dajobe> G:demo 2 [http://demo.asemantics.com/biz/isc/blue.pl|ISC blue]
09:25:24 <dc_www2003> Added comment G5.
09:25:28 <dajobe> looks for all the images for today
09:25:53 * DanC wonders if this system exports any RDF
09:26:25 <dajobe> G:demo 3 - can't find just now
09:26:26 <dc_www2003> Added comment G6.
09:27:20 <dajobe> G6:demo 3 - [http://demo.asemantics.com/biz/isc/blue.pl|Last Minute News (LMN)]
09:27:21 <dc_www2003> Replaced comment G6.
09:27:37 <dajobe> shows the channel /biz/lmn/rdf-archive/corriere.txt
09:27:42 <dajobe> [hmm, suffix]
09:27:57 <dajobe> we also describe the zie
09:28:03 <dajobe> size of images on the pages
09:28:14 <dajobe> scraping into that from theHTML
09:28:21 <dajobe> all of this is done with RDF Data Query Language (RDQL)
09:28:56 <dajobe> shows the parts of the RDQL query
09:29:00 <dajobe> can be remote or local DBs
09:29:12 <dajobe> follows the path to the info int he WHERE clause
09:29:24 <dajobe> then USING to abbreviate for the namespaces
09:29:46 <dajobe> see http://demo.asemantics.com/rdfstore/www2003/
09:29:49 <dajobe> programing with SQL
09:29:53 <dajobe> in perl, creates a DBI
09:30:03 <dajobe> then uses an RDQL query over the DBI itnerface to the rdfstore
09:30:17 <dajobe> and manipulates the DBI result set just as if it was SQL
09:30:19 <dajobe> results
09:30:22 <dajobe> lessons learned
09:30:31 <dajobe> conversion to RDF is complex but feasible especially in smaller domains
09:30:34 <dajobe> RDQL is simple
09:30:43 <dajobe> RDQL can be improved, but we want to keep it simple & efficient
09:30:50 <dajobe> - free text is important
09:30:57 <dajobe> conclusio
09:31:04 <dajobe> let's make a complex real world simple
09:31:18 <dajobe> push the coplexity to the end of the chain (in the query language) - the problems are in there
09:31:29 <dajobe> solve the problem and crunch your data in one language - RDQL
09:31:33 <dajobe> Running Code Today
09:31:40 <dajobe> where next
09:32:05 <dajobe> have rdf profiles for people, news from media, and events in calendar
09:32:11 <dajobe> these are, or can be mapped into rdf
09:32:49 <dajobe> G:[http://demo.asemantics.com/rdfstore/www2003/|RDQL demo]
09:32:49 <dc_www2003> Added comment G7.
09:32:58 <DanC> emiller: "that's one of the most sexy RDF talks I've seen"
09:33:03 <dajobe> G:[http://www.asemantics.com/download/|the code]
09:33:03 <dc_www2003> Added comment G8.
09:33:05 <dajobe> q&a
09:33:13 * DanC wonders about that slide transition; is that a powerpoint thing?
09:33:15 <dajobe> schema navigation not supported?
09:33:26 <dajobe> or recursive queyr?
09:33:54 <RSwick> [EMiller makes reference to some of the images that came up from Italian gossip papers]
09:34:26 <libby> danc, shellac says its the apple keynote
09:34:31 <libby> it's very cool
09:34:35 <dajobe> inferencing can be done in the backe end database
09:34:42 <dajobe> ontologies are going to work
09:34:48 <dajobe> we can buiold very useful stuff just with data
09:35:06 <dajobe> you can do it, it doesn't do it now
09:35:12 <dajobe> audience adds - can make it simple for now
09:35:16 <AndyS> The URL is http://www.asemantics.com/downloads.html and on to http://rdfstore.sourceforge.net
09:35:34 <dajobe> questioner replies - you need hierarchy of things & ontologies
09:35:49 <dajobe> Dirk - haven't seen a acase you need recursion in the SQL sense
09:35:54 <dajobe> the path approach is important
09:35:57 <dajobe> and useful
09:36:20 <dajobe> timbl - lots of people have looked at this as you can do it as separating inference from query
09:36:37 <Jhendler>http://www.apple.com/keynote/
09:36:38 <dc_www2003> H: http://www.apple.com/keynote/ from Jhendler
09:36:44 <dajobe> but rdql can do a simple query over an inference or closure of lower data
09:36:51 <DanC> we need a name for this point timbl/jeremy are making.
09:36:53 <dajobe> jjc - jena2 does this
09:36:58 <Jhendler> H:| Keynote slide maker
09:36:58 <dc_www2003> Titled item H.
09:37:03 <DanC> they've had to make it about 5 times so far in this conference
09:37:06 <dajobe> simple query over infered data
09:37:09 <Jhendler> H: for Apple incl. OS X
09:37:09 <dc_www2003> Added comment H1.
09:37:19 <Jhendler> H: no endorsement intended - but it looks way cool!
09:37:19 <dc_www2003> Added comment H2.
09:37:27 <dajobe> jjc - transitive closure done over lower data
09:37:47 <dajobe> emiller - adding properties to rss (alberto: yes) - can you please record them?
09:37:54 <dajobe> alberto: ACTION!
09:38:05 <dajobe> emiller - iCal, rdf caleneraring (alberto: following it)
09:38:18 <dajobe> ... talk to ical2RDF which DanC has been doing
09:38:34 <dajobe> emiller - you oculd have dfone this using other tech, what's the big gain here?
09:38:37 <AndyS>http://demo.asemantics.com/biz/buda/
09:38:37 <dc_www2003> I: http://demo.asemantics.com/biz/buda/ from AndyS
09:38:54 <AndyS> I:|Another asemantics demo
09:38:55 <dc_www2003> Titled item I.
09:38:56 <dajobe> alberto - I can add a new property and I can search it. "It's like the web"
09:39:02 <dajobe> I can add it quickly, it works
09:39:19 <dajobe> audience (@semantics) - we added a new property, immediately usable
09:39:27 <AndyS> I:Map of conference with overlay information
09:39:27 <dc_www2003> Added comment I1.
09:39:28 <dajobe> the complexity may still be there, but this stuff can work quickly
09:39:37 <dajobe> starting from simple and going as complex as you need or get paid for
09:39:56 <dajobe> kenL: what works is that the data model is rdf
09:40:43 <dajobe> dirk - even when you have a SQL db, you have written docds of the data model that is directly usable
09:40:57 <dajobe> kenL: instead of building the DB & hardcoding the schema - nightmare - you don't have to do that
09:41:03 <dajobe> l... level of indirection, the rdf model provides that
09:41:11 <dajobe> ... can ask or deal with different models easily
09:41:17 <dajobe> kenL=ken Laskey
09:41:24 <dajobe> -- end of talk
09:41:26 <libby> I:neato, he's [http://demo.asemantics.com/biz/buda/pool/albe.asemantics.com|using the rdf cal stuff]
09:41:26 <dc_www2003> Added comment I2.
09:41:27 * dajobe takes a typing break
09:43:11 <dajobe> Matt Biddulph
09:43:39 <edd> works for bbc, but hacks semweb stuff by night
09:43:47 <AndyS> G: [http://www.asemantics.com/downloads.html|asemantics download page]
09:43:47 <dc_www2003> Added comment G9.
09:43:49 <dajobe>http://www.hackdiary.com/archives/000020.html
09:43:49 <dc_www2003> J: http://www.hackdiary.com/archives/000020.html from dajobe
09:43:57 <dajobe> J:|A Semantic Web Shoebox
09:43:58 <dc_www2003> Titled item J.
09:44:09 <dajobe> J:Matt Biddulph
09:44:10 <edd> has large collection of digital photos, using RDF to index them and search
09:44:10 <dc_www2003> Added comment J1.
09:45:20 <edd>http://www.hackdiary.com/slides/www2003.pdf
09:45:21 <dc_www2003> K: http://www.hackdiary.com/slides/www2003.pdf from edd
09:45:28 <edd> K:|Matt Biddulph slides
09:45:28 <dc_www2003> Titled item K.
09:47:04 <AndyS> "don't hide the metadata"
09:47:05 <edd> puts the metadata on his site too - important to share, not just use it to build a site
09:47:16 <edd> uses RSS 1.0 as a container format for picture descriptions
09:47:26 <edd> uses DC, FOAF and Wordnet for photo annotation
09:47:37 * DanC wonders what the international gesture for "speak a bit slower, please" is
09:47:55 <dajobe> maybe it wasn't so clear but both MINDSWAP, @semantics and matt's stuff have dublin core, foaf and rss 1.0 throughout - I can see the rdf/xml detail from the front
09:48:29 <edd> K:[matt's demos|http://www.picdiary.com/www2003.html]
09:48:29 <dc_www2003> Added comment K1.
09:49:00 <DanC> hotel->building search: ooh! aaahh!
09:49:02 <edd> (sighs of pleasure from DanC)
09:49:54 <Jhendler> oooh, use case for OWL - multimedia collections - WordNet replaced by an ontology library would be powerful
09:50:14 <Jhendler> in our use case document, but he's doing it for real in a general sense
09:50:53 <DanC> yup, let's put that one in 2001/sw/WebOnt/rqim
09:51:25 <DanC> though I suppose it's really just an RDF/RDFS use case... any particular way to exploit OWL, Jhendler?
09:52:21 <dajobe> "you hit a picture, and it's suyrrounded by the related pictures in semantic space"
09:52:47 <DanC> wordnet hyper/hyponym, geo, etc.
09:52:52 <dajobe> ^- that's what he hopes to do, improving the interface he showed
09:53:30 <DanC> using redland
09:53:31 <edd> uses redland (go dajobe!)
09:53:39 * dajobe :)
09:53:41 <Jhendler> DanC - exploitation of OWL would be (i) able to shift into more powerful relationship models (i.e. could wander into Guus' "great art" browser)
09:53:58 <DanC> more powerful... like what?
09:54:05 <AndyS> Example JimH?
09:54:11 <Jhendler> (ii) could find other peoples representations to ap yours against (i.e. you pictures of fish map to his pictures of scuba diving)
09:54:45 * edd applauds interesting UIs for this stuff
09:54:45 <Jhendler> Andy - sure, supposing you wanted to see only "tall" buildings - keywords won't link that to "skyscraper" nor will wordnet
09:54:47 <dajobe> shows the dynamic rss behind the scrreens being used via the phil polaroid interface
09:54:59 <AndyS> Connecting shoeboxes - great example
09:54:59 <DanC> both (i) and (ii) look like RDF/RDFs, Jhendler. I wonder how to exploit the additional OWL stuff.
09:55:24 <Jhendler> but if you wandered into an "architecture" ontology it could organize your terms (linked back to wordent)
09:55:28 * AndyS is asking same as DanC
09:55:50 <DanC> what terms from OWL that are not in RDF/RDFS would your example use, Jhendler?
09:55:54 <Jhendler> OWL would allow things like only looking for "females" when the person said, girl, woman or "muchacha"
09:56:16 <DanC> RDFS allows that, no, Jhendler?
09:56:21 <edd> female humans, i hope
09:56:26 <dom> !
09:56:29 <dajobe> mattb mentions emitting rss1.0 needing to do it via a profile/template since couldn't control the output too much
09:56:34 <Jhendler> so I can imagine that one would end up in a space, the ontology could be rendered as a "form" and we could further subdivide
09:57:05 <Jhendler> I was thinking of being able to use owl "axioms" like female disjoint from male" etc
09:57:07 <DanC> what owl terms, Jhendler?
09:57:29 <DanC> ok, disjoint is in owl but not RDFs.
09:57:33 <Jhendler> so what would happen is I would specify a couple of properties, OWL would classify my instance to some class, and then retrieve in that class
09:57:52 <Jhendler> the use of this kind of classification is a property of the many terms in OWL DL
09:58:50 <Jhendler> uses cardinality (optional v. required), restrictions, disjointness, uniqueness (fins same person as one in this picture) and mappings
09:59:01 * DanC can sorta see how to use owl:onProperty/owl:hasValue, but can't quite see the concrete example
09:59:08 <Jhendler> (i.e. the Scuba to Fish thing would be owl:equivalentTo)
09:59:09 <edd> now onto location based annotation of items
09:59:36 <edd> uses spacenamespace ontology for locations
10:00:03 <AndyS> "same as" (owl:sameIndvidiualAs) is something we come across quite a lot
10:00:05 <Jhendler> ontology defined as RDF-Schema (err VDL) in this example?
10:00:30 * edd thinks so. owl might be used. dajobe do you know?
10:00:43 * DanC notes owl:sameIndividualAs has a synonym, owl:sameAs
10:00:47 <dajobe> yes, for smooshing
10:00:58 <edd>http://space.frot.org/
10:00:59 <dc_www2003> L: http://space.frot.org/ from edd
10:01:04 <edd> L:|spacenamespace
10:01:05 <dc_www2003> Titled item L.
10:01:22 <AndyS> What's the relationship to owl:sameClassAs? (or have I misremembered?)
10:01:31 <edd> L:"it is a semantic web project; it provides a scheme for semantic web identification of places via unique uris"
10:01:31 <dc_www2003> Added comment L1.
10:01:34 <Jhendler> dajobe - cool, will look into that for my implementation status report
10:01:39 <DanC> owl:sameAs rdfs:subPropertyOf owl:sameClassAs.
10:01:52 * AndyS exposes lack of time to understand owl :-(
10:02:06 <dajobe> er, I meant he could use owl, I don't think he does right now
10:02:29 <DanC> sameAs means exactly identical; sameClassAs just means they have the same members (not, necessarily, the same label/comments/colors/etc.)
10:03:10 <Jhendler> dajobe - not could should, but then I'm biased :->
10:03:19 <AndyS> Got it - thx DanC
10:04:00 <dajobe> Jhendler: sure, should - I think he knows that
10:04:05 <Jhendler> andyS, not surprising - takes a bit to learn each lang.
10:04:25 <AndyS> Time, time
10:04:50 <AndyS> Is owl:sameAs in OWL-lite?
10:04:52 * Jhendler bit of an OWL bigot these days, but also know that many apps don't need the power YET - so I consider every RDFS app to be one that will migrate to OWL eventually - it only takes a tiny bit to start adding the OWL vocab
10:05:21 * DanC hasn't committed the owl-lite restrictions to (mental) memory; looks up owl:sameAs...
10:05:31 <dajobe> I think mattb's probably waiting for tool support; he juis tmentioned having to pre-calculate the subsumptions sinc ehe had no inference in what he was using (jena I think)
10:06:11 <AndyS> Jena2 has inference. Lets us know if it does what you require!
10:06:14 <Jhendler> yeah - one thing that might be nice in OWL is having that happen "for free" -- each new photo would automatically get indexed in the right places
10:06:16 * em-lap likes the term 'semweb-space'
10:06:34 <DanC> cwm can do that sort of pre-caclulation, if you don't need it to be too quick
10:06:52 * Jhendler +1 to em-lap
10:06:58 <DanC> euler can probably do the backward-chaining in interactive time
10:07:14 <em-lap> [/me hears need again for open geo space vocab and instance data]
10:07:49 <em-lap> [/me takes action item to follow up with getty and TGN]
10:07:58 <dajobe> "boinging down stuff to get a sense of nearness"
10:08:03 <dajobe> boiling
10:10:05 <edd> talking about difficulty of annotation interface
10:10:11 <edd> produced a bot as a tool to annotate
10:10:22 * DanC noodles on voice interfaces to photo annotation...
10:10:31 <edd> the bot would be stateful, you annotate via 'conversation'
10:10:41 <JibberJim> (/me assumes mattb is demoing his pic stuff)
10:10:54 <edd> JibberJim: yes
10:10:58 * Jhendler thinks the annotation is a good use case for OWL, see earlier discussion)
10:11:20 * Jhendler i.e. Owl can map to forms easily, and the restrictions help in filling them out correctly
10:11:33 <dajobe> "data really really portable and malleable"
10:11:41 * Jhendler thinks everything is a good use case for OWL :-> foo - I'm getting to be an OWL bigot. apologies to all...
10:12:13 * AndyS thinks it would be good if people used OWL by default.
10:13:03 * AndyS (but what level?)
10:13:57 <dajobe> really simple query thing makes you go a lot faster
10:14:02 * edd notes we need more docs. anyone want to write me an OWL primer and quick reference for xml.com?
10:14:52 * Jhendler AndyS - depends on needs, but suspect Lite has most of the vocabulary we want
10:15:49 * AndyS would suggest start low (instead of RDFS) to leave growth later as data grows
10:16:47 * DanC is finding it surprisingly hard to figure out if owl:sameAs is in OWL Lite
10:17:07 <DanC> OWL primer... the OWL guide doesn't work?
10:17:13 <dajobe> table at the start of owl ref?
10:17:20 <dajobe> er, s&as
10:18:08 * Jhendler equivalency is in Owl Lite - I think we forgot to include sameAs in the list - but it is in Lite as far as I can tell
10:18:23 <AndyS> Value is the simple "write your ontology terms down" not just the inference
10:18:27 * timbl and Danc and Sandro expanded teh N3 primer a lot recently for tutorial at thisconference.
10:19:59 <edd> mentions problem of RSS overcrawling, need way of sharing and distributing "what's new on the semantic web" info that will scale\
10:20:25 * AndyS would like to see a "standard" N3 - encourage people to write it (c.f. std query)
10:20:55 * AndyS wonders what happened to the effort on this
10:21:14 <DanC> really, Andy? why should people be encouraged to write N3? I think people should export RDF/XML; N3 is just a poor-man's user interface.
10:21:26 <timbl> Standrda in what sense?
10:21:33 <timbl> A NOTE?
10:21:39 <timbl> A WD on rec track?
10:21:53 <AndyS> Value of std is stable + documented so NOTE is perfect
10:22:27 <AndyS> DanC - exactly - collects small data items We have seen this done today at least twice
10:22:39 * dajobe knows AndyS has an annotated grammar with some gotchas
10:22:40 <AndyS> UI's are better but take time!
10:23:00 * AndyS just tracks cwm by testing when he has time
10:23:12 * AndyS is behind at the moment
10:24:01 * AndyS is not standardising N3
10:24:19 <DanC> oops; actually, I'm only using Zope because it plays nicely with iPhoto
10:25:05 * edd laments poor general webdav interop on linux.
10:25:57 <DanC> poor webdav interop? gee... WorksForMe; I upload with iPhoto to zope running on linux, then use cadaver to grab and edit the .html, then switch to Amaya. I'm FatDumbAndHappy when it comes to webdav and linux.
10:26:56 <DanC> BBC is publicly funded?
10:27:03 <edd> Yes.
10:27:08 * DanC has been in some kind of hole or something
10:27:11 <edd> TV owners pay a licence fee each year.
10:27:13 <RSwick> by taxes, err, licenses
10:27:28 <RSwick> "tv stamps"
10:27:28 <JibberJim> It's "the unique way the BBC is funded"
10:27:58 <libby> it's about 100 quid a year, bit more
10:28:12 <DanC> re TGN and the like... and licensing issues... "Evangelism or perl + duct tape?" http://www.w3.org/2003/Talks/0520-www-tf1-d3-travel/slide11-0.html
10:28:14 <AndyS> per house
10:28:22 <JibberJim> 140 per house with TV - 5 quid discount if you're blind IIRC.
10:32:47 <Jhendler> looks through tutorial -- really nice - we are going to teach students python, so we should have a team of cwm hackers by end of next term.
10:33:04 <DanC> python tutorial? yes, it rocks.
10:33:42 * DanC finds a "team of cwm hackers" somewhat frightening, but perhaps in a good way
10:33:58 <Jhendler> DanC - jumped into middle - meant N3 tutorial, but +1 on python tutorial as well
10:35:41 * Jhendler been looking for a name for the lab -- anyone know another word with no vowel except "w"?
10:36:07 * JibberJim shoulda played more scrabble....
10:36:41 * Jhendler hey - I've memorized all the words with q but no u...
11:00:13 <zool> how did matt's talk go
11:00:20 * zool sees the slides linked from the chump
11:04:55 <zool> hrm i should fix up my ontology
11:05:20 <zool> in a bit of a state since i let the bots start rewriting it
11:07:33 <JibberJim> you have your bots write the ontology?
11:07:43 <JibberJim> clever bots.
11:09:40 <zool> well, i aspire to... the space navigating bot picks up new classes of things, and sends them via jabber to an ontology-compiling bot... that has a little voting system
11:10:32 <zool> been given the idea of using wordnet - or TAP, i guess - ontologies to help 'pre-place', connect new classes into my little taxonomy of space
11:11:39 <zool> i am not actually seeing the TAP ontology for places expressed in OWL on the tap site though, just all the samples in the kb...
11:11:40 * JibberJim started a places vocab for translating my GPS waypoint names to a Class.
11:11:47 <JibberJim> - http://jibbering.com/vocabs/places
11:13:11 <zool> that's nice, i should put the equivalences in my one - http://space.frot.org/rdf/space.owl being the last version - quite possibly invalid/odd
11:13:46 <zool> (trying to collapse that in back into the main space namespace now, wondering if that makes sense)
11:13:49 * zool sorry channel drift
11:18:07 * JibberJim should probably just use yours actually, but I didn't know about it when I wrote my GPS data converter.
11:19:05 <zool> well, i'd like to fix the ontobot and open up the ontology for public frotting about
11:55:24 <afs> afs is now known as AndyS
12:02:00 <DanC> --- reconvene after lunch
12:02:28 <DanC>http://kaon.semanticweb.org/
12:02:30 <dc_www2003> M: http://kaon.semanticweb.org/ from DanC
12:02:47 <DanC> M:|KArlsruhe ONtology Management Tool Suite
12:02:48 <dc_www2003> Titled item M.
12:03:00 <DanC> M:presentation (now) at devday
12:03:00 <dc_www2003> Added comment M1.
12:03:07 <DanC> M:source code available; LGPL
12:03:08 <dc_www2003> Added comment M2.
12:07:12 * DanC doesn't mind folks that constrain their use of RDF, as long as they're up front about it
12:08:05 <sandro> Do you think their data published using rdfs:range will really be interoperable with that of someone who uses rdfs:range the other way?
12:08:40 <DanC> yes; am I misunderstanding something?
12:08:40 <sandro> "yes, we make the unique names assumption, but I'll get back to that."
12:09:40 <DanC> what might not work is other people's data in their tool
12:09:44 <sandro> I guess it depends when and how you close the world.
12:10:03 <DanC> I don't see any closed-world stuff yet.
12:10:54 <sandro> Hm.
12:19:02 <sandro> using wordnet as a kind of loose upper ontology, to help you find the real ontology you want.
12:24:27 * Jhendler thanks sandro -- that last line (using wordnet...) has just broken a logjam on a project I've been breaking my head over
12:25:36 <sandro> attribution to KAON of course.
12:33:42 * sandro sees the duality between set operations and boolean operations yet again, with again one approach being argued as better....
12:34:12 * AndyS can't see why the specific syntax used can't be compiled to a single conjunctive expression - maybe its the architecture precludes this
12:34:54 <sandro> I think Grosof/Horrocks DLP algorithm should convert between the two.
12:36:11 <AndyS> Got it. Having a syntax that is task-specific is a useful thing.
12:36:41 <sandro> yeah. so is having a standard syntax. :-)
12:37:48 <sandro> [foo] means the class foo, <foo> means the property foo
12:37:56 <AndyS> I would not like query in RDQL in their world. The ontology/data split is a user concept that is natural
12:38:10 <AndyS> c.f. RQL
12:38:47 <sandro> yeah. two standards, maybe? :-)
12:39:08 <AndyS> Typed expressiosn being used to see difference usages
12:39:28 <RSwick> Andy, that split is perhaps "natural" mostly to 'users' who are prohibited from talking _about_ ontologies
12:40:05 <AndyS> True.
12:40:14 <RSwick> ... i.e. they live in a very constrained world, don't realize there are other worlds ;)
12:40:48 <sandro> Ian: This is not a motivating example for having classes and properties and instances with the same names. This comes from loose natural language use.
12:43:45 <sandro> speaker mentions the *other* cwm (Common Warehouse Model)
12:45:36 <sandro> Jeremy suggests ontologies might include a mailto for subscribing to the version-update list.
12:47:01 <sandro> woo hoo! does that mean we have a haystack release?
12:47:40 <AndyS> Haystack will open source (if it hasn't already) soon - I heard by 1 June
12:47:43 * timbl holds his breath
12:47:49 <dajobe> lol
12:48:05 * dajobe checks, he seems to be still breathing
12:48:39 <AndyS>http://haystack.lcs.mit.edu/
12:48:39 <dc_www2003> N: http://haystack.lcs.mit.edu/ from AndyS
12:48:44 <dajobe> Haystack - friendly semantic web, Dennis Quan
12:48:55 <AndyS> N:|Haystack
12:48:56 <dc_www2003> Titled item N.
12:49:29 <AndyS> Download area says 1 June 2003
12:49:57 <AndyS> "Can't believe everything you read on the web" - like release dates?
12:50:02 <mattb> i was talking to one of the haystack guys at lunch and he says he'd rather they waited another month or so, it's still not ready to show
12:50:44 <dajobe> lol
12:50:57 <dajobe> if I'd had that attitude, I'd still not have shipped redland ;)
12:51:03 <AndyS> mattb - any idea why?
12:51:15 <dajobe> ... keep in touch with friends not their homepages"
12:51:35 <dajobe> "I cannot bookmark a gene"
12:51:37 <mattb> i think he wants more refactoring done, plus the hardware requirements are high and some worries about it being underwhelming
12:51:48 <dajobe> hmm
12:51:51 <mattb> underwhelming because of performance, i mean
12:51:53 <mattb> not features
12:52:00 <dajobe> metadata - relating people, things, and tracking things => rdf
12:52:09 <AndyS> Longer the wait, the greater the expectations.
12:52:12 <dajobe> yes
12:52:25 * dajobe shrugs, it's still ok
12:52:58 <dajobe> making things tangible - copuing data between places, rather than the data having a "place"
12:53:09 <dajobe> The movie is important not the movie info sites
12:53:25 <dajobe> when you do have first class data - data centric - you can start from there
12:53:33 <dajobe> and operate on it, drag drop it
12:53:48 <dajobe> right tools for the job - calendar written as a graph usong graphviz
12:53:55 <dajobe> example as it is wrong
12:53:57 <dajobe> ;)
12:54:19 <DanC> this guy has the presentation gene, clearly
12:54:26 <dajobe> yeah
12:54:41 <dajobe> smileys in a slideshow
12:54:58 <dajobe> me me me - stuff for me, importnant to me, my way, metadata for me
12:55:17 <dajobe> "or you, of course"
12:55:29 <dajobe> haystack - general info environment built on rdf
12:55:43 <dajobe> automatically records metdata from users as they work
12:55:56 <dajobe> uses rdf semantics for presentation and commands for the user
12:56:15 <dajobe> user interface which can consume and produce rdf, is an interface for the semantic web
12:56:18 <dajobe> haystack demo
12:56:26 <dajobe> "looks superficially like a portal"
12:56:34 <dajobe> bookmarks, favourites, tasks, buddy lists
12:56:42 <dajobe> email inbox, news feed, weather, ...
12:56:58 <dajobe> right hand side has some tools bars
12:57:24 <dajobe> he does a demo
12:57:35 <dajobe> email with attached todos
12:57:56 <dajobe> and a bookmark for a travel web ste
12:58:12 <dajobe> another email with a conf paper ptr
12:58:25 <dajobe> he clicks on the conference in active tasks to indicate it's associated
12:58:41 <dajobe> toolbar on the right changed to match the type of thjing - an event
12:58:43 <DanC> I do this with my web browser: WindowPerTask, from the smalltalk days.
12:59:06 <dajobe> createas a category 'things to do and see' for the conference
12:59:07 <DanC> but it would be cool to get this sort of support.
12:59:29 <dajobe> he selects the doc which is a urn:lsid:....
12:59:49 * DanC wonders why not http
12:59:55 <dajobe> created a new urn, web service & retrieval service based on web services that gets metadata about it
13:00:01 <DanC> i.e. http://lsid.org/...
13:00:05 <dajobe> clicks on it to get a bynch of article properties
13:00:22 * sandro guesses it's because like 99% of web publishers, they dont believe http URIs can be persistent.
13:00:28 <dajobe> the toolbar changes to match the type of thing
13:00:40 <dajobe> "to match the huge amount of things the system can do"
13:00:44 <dajobe> [hmm]
13:01:10 <DanC> 99 out of the top 100 web publishers know they need to pay attention to persistence.
13:01:24 <dajobe> he selects a related category then switch es the view to a graph of context of the article
13:01:34 <dajobe> such as relating to the collections and the authors
13:01:48 <dajobe> clicks on "Jim Brown (avaiible for chat)"
13:02:05 <dajobe> right click to get actions on the object
13:02:11 <dajobe> - can be done on any object -
13:02:17 <dajobe> he does so, to remember to contact him
13:02:19 <dajobe> annotations
13:02:30 <dajobe> types a quick one on the paper
13:02:50 <dajobe> back to home page
13:02:53 <dajobe> spam mail
13:03:06 * DanC feels like he's at a hypertext conference, watching somebody from Brown university navigate a really, really cool user interface that's much like emacs: steep learning curve, but once you master it, you're zillions times more efficient at all sorts of stuff
13:03:39 <dajobe> yeah
13:03:53 <dajobe> more mail management, categorising
13:04:03 <dajobe> he ticks mail, lit refs as done ina ctive tasks
13:04:08 <dajobe> move to flight to a conf
13:04:45 <dajobe> system suggest's related activities of planning travel, he acks
13:04:57 <dajobe> it offers more features for the user - contacts at the dest, weather
13:05:23 * DanC wonders if it knows about "goals" and such
13:05:28 <sandro> I love Dennis' demo humor.
13:05:36 <DanC> "ipedia" ;-)
13:05:39 <DanC> impedia
13:05:50 <dajobe> he uses the travel site web service
13:05:55 <dajobe> shows the rdf
13:06:01 <dajobe> adds it to the system
13:06:06 <JibberJim> travel site web service?
13:06:11 <dajobe> fake one
13:06:21 <dajobe> "when wsdl people work out rdf is better"
13:06:23 <dajobe> s/work/figure/
13:06:25 <JibberJim> ah, /me is working on real one..
13:06:33 <dajobe> heh
13:06:49 <dajobe> books flight, it add's to his todo list
13:07:18 <dajobe> missing info is who pays, so he goes to the org chart
13:07:30 <dajobe> which is a list, but he picks as a graph, finds his boss
13:08:05 <dajobe> more form filling, done
13:08:19 <dajobe> his ws returns immediately, being local
13:08:28 <dajobe> gets a receipt with ticket in email
13:08:42 <dajobe> drags it to things to see and do
13:08:48 <dajobe> and adds it to expenses
13:09:10 <dajobe> ticks planning travel ,done
13:09:14 * DanC considers an evil question: does Dennis really use this for his real email yet?
13:09:19 <dajobe> selects the person at the dest, to contact
13:09:39 <dajobe> lots of personal info
13:09:50 <dajobe> in this case, the person is online too
13:10:00 <dajobe> buddly list, quick chat box
13:10:33 <dajobe> friend replies to lunch chat request yes
13:10:40 <dajobe> with an attached event
13:11:31 <dajobe> he drags the lunch to things to see and do
13:11:53 <dajobe> he browses that as a last, changes view to calendar
13:12:08 <sandro> "view as calendar" "view as tree" wow. wow.
13:12:28 <dajobe> he adds to calendar
13:12:30 <dajobe> end demo
13:13:05 <dajobe> users - users info is being stored locally and displayed on user prefs
13:13:05 * yod didn't see demo... but that sounds like a quite obsessive way to run one's life :)
13:13:12 <sandro> but there is no "calendar", just a pile of triples, some of which have time relations.
13:13:12 <DanC> ok, now that he's whet our appetite, just give us the download pointer and stand back for the round of applause.
13:13:15 <dajobe> developers - get ontolkogies for presentation and user info, more metadata
13:13:22 <dajobe> DanC: ha ha
13:13:55 <dajobe> haystack uses basic user information paradigms [basic? familiar,...]
13:14:10 <dajobe> context of type of info and current task of the user indicates relevance
13:14:28 <dajobe> histories, bookmark collections are tied to contexts
13:14:34 <AndyS> N:[http://haystack.lcs.mit.edu/downloads.html|Download page]
13:14:34 <dc_www2003> Added comment N1.
13:15:02 <dajobe> collections and taxonomie - albums, listings, hierarcies, sets, folders - differ only in presentation style
13:15:05 <dajobe> [really??]
13:15:17 <AndyS> N: Release due 1 June 2003
13:15:17 <dc_www2003> Added comment N2.
13:15:39 * mattb notes that it's coded in java and uses the ibm SWT toolkit for native windows/gtk2/etc widgets (like eclipse)
13:15:43 <dajobe> "people live out evamil, cos they like the list format for dealing with things todo" and don't like the other model
13:15:53 <dajobe> the Haystaack GUI manual is online already
13:15:55 <dajobe> with pics
13:16:01 <dajobe> I read it about 2 weeks ago
13:16:21 <dajobe> collaboration - comms across multipel channels, IM, email, p2p. metadata and "expertese" can besent
13:16:42 <AndyS> N:[http://haystack.lcs.mit.edu/documentation.html|Documents]
13:16:42 <dc_www2003> Added comment N3.
13:16:45 * DanC wonders if the chandler folks have seen any of this
13:16:52 <DanC> comment N3 ;-)
13:16:53 * AndyS thanks dajobe for pointing that out
13:16:57 * sandro wondered that too
13:17:02 <dajobe> DanC: I wondered about that, seems if Haystack had been public....
13:17:23 <dajobe> developers bbuilding blocks, see N3 for adenine, ozone, etc.
13:17:33 <dajobe> rather chump item N3, not the language - lol
13:17:54 <dajobe> adenine - oo code in RF, interfaceing with rdf sores
13:17:59 <dajobe> stores
13:18:03 <dajobe> can send code
13:18:04 * DanC wonders what prize Andy should get for being caller number N3
13:18:48 <dajobe> reference to lisp
13:19:25 <dajobe> ozone UI language declaritive with primitives similar to html
13:19:38 * AndyS waits for M4 comment (old joke)
13:19:59 * dajobe waits for N3-lite comment
13:20:17 <dajobe> views - components to display types of objects in certain ways
13:20:31 <dajobe> such as a calendar, sumamry view, ...
13:21:12 <dajobe> property editor view - key-value pair editor
13:21:17 <dajobe> forms are fundamental
13:21:23 <DanC> these guys seem to be fearless; they've build a programming language, a component document architecture, a GUI designer, what else?
13:21:25 <dajobe> values are views
13:21:27 <DanC> built
13:21:32 <dajobe> yeah
13:21:39 <dajobe> chance of everyone switching to all of that?
13:21:49 <dajobe> see smalltalk, oberon, squeak, ...
13:21:57 * dajobe ducks, sees no bijan
13:22:20 <dajobe> graph editor views - emphasisingconnectoons between objects, at a certain level of granularity
13:22:31 <sandro> chances of everyone switching to XSLT? about the same.
13:23:21 <DanC> well, in the W3C track there was a session about the next gen web client; I've seen SVG+javascript used as a new platform. I can believe a new platform will be born in the next few years.
13:23:31 <dajobe> yes, but not haystack
13:23:37 <AndyS> They do highlight each of the important technology pieces - it paints a big picture and shows it can work
13:23:42 <dajobe> it might show pieces... yeah
13:23:48 <DanC> why not haystack?
13:24:04 <dajobe> their own gui elements, object model, language
13:24:38 <dajobe> quan - outlines system wiorks
13:24:40 <DanC> the gui is proven, no? I've seen that Java IDE endorsed highly
13:24:46 <dajobe> quan - on to bioinformatics demo
13:24:56 <DanC> the object model is RDF. WorksForMe.
13:25:11 <DanC> the new language: I don't know enough about it to say that it's no good.
13:25:37 <dajobe> he uses xslt to put things 'into rdf'
13:25:50 <DanC> er... so do I.
13:26:03 <dajobe> says that the info would take 20 windows do normally display the aggreaated info
13:26:06 <DanC> eclipse; that's the IDE.
13:26:20 <edd> java 1.4.. :( only 1.3 supported on ppclinux so far.
13:26:20 <dajobe> impelemtnation java 2 v1.4 - runs undwer windows fine, linux so-so, mac osx - not tried
13:26:29 <mattb> it's a nice widget set - maps natively to windows and gtk2 widgets
13:26:35 <dajobe> release - "very intiial pre-release" 1 June 2003
13:26:37 <mattb> unlike ugly one-size-fits-all java swing guis
13:26:47 <dajobe> "can't make any promises about stability or data model"
13:26:51 <dajobe> "by august, better"
13:27:11 <dajobe> we built a custome rdf store since gui ind atat model, was very taxing, needed perf
13:27:13 <dajobe> conclusion
13:27:21 <dajobe> 1) haystack is a user interface fo rthe semweb
13:27:37 * DanC noodles on using this for managing W3C issues lists etc.
13:27:44 <dajobe> - visualizes, ui toolkit, point & click creat and maniupluate, ui metadata, access to sem/ws/grid services
13:27:51 <dajobe> 2) makes the semweb useful
13:28:01 <dajobe> for everyday use
13:28:22 <dajobe> - browse, organize semweb and take advantage of user's context when retrieving info
13:28:50 <dajobe> refernce to context-based search, will talk to TAP
13:28:56 <dajobe> (already using TAP KB)
13:29:07 <dajobe> future work - fromd aml+oil to owl, more xsd datatypes, some owl concepts in UI
13:29:13 <dajobe> - scurity and privacy
13:29:27 <dajobe> [hmm, not done that yet?!]
13:29:27 * DanC notes use of DAML+OIL in haystack UI, headed for OWL
13:29:38 <dajobe> - ontology conversion
13:29:43 <dajobe> - better ws hookups
13:29:44 <timbl> q - how manage storage backuop and access
13:29:55 <DanC> yeah, tim
13:30:09 <dajobe> q&a
13:30:26 <DanC> hmm... along with chandler, it reminds me of sherlock
13:33:06 <DanC> Q: done user testing?
13:33:07 <dc_www2003> Label Q not found.
13:33:45 * edd packs up, prepares to head off
13:33:48 <DanC> Dennis: yes, I have a user study going on, for a technically-competent user community
13:34:20 <DanC> Dennis: my target is that if somebody can use outlook, they'll be able to use this
13:36:36 <DanC> Martin: I see the support for context and switching, but people switch context without, perhaps, telling the machine. Do you support that?
13:37:08 <DanC> Dennis: these days, people switch contexts by switching applications...
13:37:49 <DanC> TV/Dennis doesn't buy it; the web browser spans contexts
13:38:23 <DanC> Dennis: yes, open source release [web site says June or July or some such]
14:08:46 <RalhS> RalhS is now known as RSwick
14:10:48 * DanC crosses fingers and hopes for TAP to start using the Joseki web api
14:10:51 <DanC> i.e. GET
14:10:53 <dajobe> edd has left the building ...
14:10:55 <sandro> Andy: Joseki tries to be minimal, not ideal
14:11:35 <sandro> andy: publisher does not know what consumer is going to do with the data !
14:11:44 <DanC> joseki is the natural peer to something like haystack
14:11:44 <dajobe>http://www.joseki.org/
14:11:44 <dc_www2003> O: http://www.joseki.org/ from dajobe
14:11:49 <sandro> andy: different toolkits at publisher and consumer sites
14:12:40 <DanC> ooh! email adapter... I wrote one of those...
14:12:42 <dajobe> O:|Joseki server for publishing RDF models, Andy Seaborne
14:12:43 <dc_www2003> Titled item O.
14:13:17 <DanC> is there a chump item for the presentation yet?
14:13:55 <DanC> O:presentation slide "example: meeting assitant" slide has an email adapter bubble. I wrote one of those: [http://www.w3.org/2000/04/maillog2rdf/mid_proxy.py|mid_proxy.py]
14:13:55 <dc_www2003> Added comment O1.
14:14:46 <dajobe> urls for rdf graphs
14:15:01 <dajobe> experimentigng at level of granularity of web api for semweb
14:15:03 <DanC> and for queries! (take note, folks doing GetData over SOAP)
14:17:05 <dajobe> oops
14:17:29 * DanC is curious about update
14:17:56 <DanC> hmm... mutlipart/form-data seems like an easy way to send 2 graphs in a POST/update request
14:20:07 <DanC> the "subgraph covered by the query" idea seems like a cool way to distribute the processing.
14:21:11 * DanC finds Andy's design intiutions seem to match mine quite nicely ;-)
14:23:03 * DanC can't hear; should I interrupt? anybody else who can't hear?
14:23:41 * sandro wondered how anyone in the back of the room could hear.
14:24:29 * DanC notes zero eye contact between the front of the room and the back; Andy's making eye-contact with the front row, and folks in the back are staring at their machines
14:24:40 * mattb can't hear much
14:28:23 * RSwick wonders why people who can't hear won't move up to fill all the empty chairs closer to the speaker. He shouldn't have to shout.
14:29:17 <mattb> fair comment
14:30:07 * timbl notes TV Ramin is quite close to the front. Were he closer, he would be even closer t the projector.
14:30:33 * danbri reappears from rdfs hacking w/ brian
14:31:10 <danbri> mattb, bummed i missed your talk... did you have fun?
14:32:13 <mattb> danbri: yes, lots of interesting comments, questions and suggestions afterwards
14:32:44 <danbri> excellent :)
14:32:51 <danbri> its nice work
14:35:29 <DanC> Norm uses N3 to configure his jpegrdf doodad too. (hmm... do we owe notice to www-rdf-config?)
14:40:08 <dajobe> promise that the slides will be online
14:44:03 <DanC> I wonder if anybody's glued the euler backward-chaining engine into the Jena API
14:44:40 <danbri> that would be v cool
14:45:38 <DanC> I bet timbl will offer an N3 note when Joseki supports N3 queries (in the URL, i.e. in GET)
14:46:10 <DanC> vcard:* is like log:uri.log:startsWith
14:49:29 <DanC> copy what?
14:51:29 * timbl (coiipy the uri in his slide)
14:53:04 <timbl> The question of graph-walking enough to identify a bnode isinteretsing.
14:53:26 <timbl> AndyS pointed to Alex's paper.
14:53:39 <timbl> (Alex doing finals at UCL)
14:53:46 <danbri> jeremy's paper too i guess
14:53:51 <danbri> (re graph match)
14:53:56 <DanC> such that if nodeX in G denotes A and nodeX in G denotes B, then we can conclude A=B, right?
14:54:06 * timbl hears horn which normally means plenary session ?!?
14:54:19 <danbri> theres a wedding party upstairs, it might have different semantics now
14:54:56 <AndyS>http://www1.bcs.org.uk/DocsRepository/03700/3772/barnell.htm
14:54:56 <dc_www2003> P: http://www1.bcs.org.uk/DocsRepository/03700/3772/barnell.htm from AndyS
14:55:06 <AndyS> P:|RDFObjects
14:55:07 <dc_www2003> Titled item P.
14:55:19 <AndyS> A paper by Alex Barnell
14:55:31 <AndyS> P:A paper by Alex Barnell
14:55:31 <dc_www2003> Added comment P1.
14:55:50 * em-lap notes redland plug again
14:55:54 <danbri> prev annotea presentations: http://www.w3.org/2001/Annotea/User/Papers.html (pre-bookmarks)
14:56:08 * timbl concludes it is something in the auditorium, which is now gated off.
14:56:19 * danbri would like to see raptor in mozilla instead of their crufty old rdf parser
14:56:31 <mattb> +1
14:57:51 <em-lap> err.. shut the door tim
14:58:01 <em-lap> thanks :)
14:58:04 * danbri would also like to see a compare/contrast between mozilla rdf bookmarks vocab, vs annotea new stuff, vs weblogging and rss markup
15:00:13 <DanC> mozilla rdf bookmarks vocab? pointer?
15:01:19 <alberto> how is the RDF update in amaya/redland works ? add/remove triple ?
15:01:49 <alberto> is there any issue with bNodes when changing something down in a graph ?
15:02:36 <dajobe> update is presently 1) remove triple 2) add new one
15:02:50 <dajobe> but I asked jose and he mostly needs triple.setObject()
15:03:33 <alberto> dajobe, clear -it must fit at the same bNode "parsing position" (rdf:nodeID) right?
15:03:58 <dajobe> ?
15:04:07 <alberto> we are having similar problems and contexts seems helping in there - glob triples per object or something
15:04:54 <DanC> "conclusions" slide: split in 2, please.
15:05:11 <alberto> dajobe, I mean when you zap a triple with a bNode in you need to insert it back at the "same" position - or am I missing somehting?
15:05:33 <dajobe> oh, I see
15:05:40 <dajobe> yes
15:06:52 <DanC> cwm has a way of cheating... there's some flag where you can have it assign nodeids based on the line and character where it was parsing.
15:08:06 <alberto> right, good hack DanC - even if is not the general solution
15:08:59 <dajobe> Isa Viz - Emmanuel Pietriga
15:09:03 <dajobe> GSS - Graph Style Sheets
15:09:10 <dajobe> based on Jena2
15:09:25 <dajobe> import/export modules, can for e.g. use sesame stores
15:09:26 * DanC wants to compare GSS with "circles and arrows using stylesheet rules" http://www.w3.org/2001/02pd/
15:09:36 <alberto> I found "contexts" useful to select all triples of an object I want to update - zap/remove them and then insert the new updated object back in the same or different context
15:10:08 <dajobe> redland has model methods for that - model.delete_triples_in_context(context_node)
15:10:16 <DanC>http://www.w3.org/2001/11/IsaViz/graphstylesheets/gssmanual.html
15:10:17 <dc_www2003> Q: http://www.w3.org/2001/11/IsaViz/graphstylesheets/gssmanual.html from DanC
15:10:27 <DanC> Q:|Graph Stylesheets (GSS) in IsaViz
15:10:27 <dc_www2003> Titled item Q.
15:10:35 <dajobe> ooh, svg slides
15:10:56 <dajobe> demoing isaviz 2.0 alpha
15:11:07 <alberto> yeah dajobe, we have remove($triple,$context) as well as add/search ones
15:11:22 <RSwick> Emmanuel's SVG slideset at http://www.w3.org/People/Emmanuel/www2003/all.svg
15:11:37 <dajobe>http://www.w3.org/People/Emmanuel/www2003/all.svg
15:11:37 <dc_www2003> R: http://www.w3.org/People/Emmanuel/www2003/all.svg from dajobe
15:12:19 <dajobe> R:|IsaViz slides - Emmanuel Pietriga (SVG)
15:12:19 <dc_www2003> Titled item R.
15:12:48 <dajobe> wee! that was pretty fast for java svg
15:12:59 <dajobe> (I guess it's not svg in the app)
15:13:20 <em-lap> no its not svg in the app - but you can export this in svg
15:13:20 <RSwick> no, IsaViz does not render using SVG
15:13:42 * danbri likes the svg slides... wonders what tool made them
15:13:47 <DanC> what did he hide?
15:14:03 <RSwick> a bunch of "less interesting" arcs in the bookmark.rdf
15:14:25 <DanC> er.. how did he tell the machine what to hide? I'm having trouble hearing
15:14:38 <RSwick> uses CSS property names & values -- expressed in RDF
15:14:52 * Jhendler wishes he would talk a tad louder (and a bit slower)
15:14:54 <RSwick> he's loading gss files and applying them
15:15:02 <RSwick> pre-prepared gss files
15:15:24 <timbl> TR = technical report
15:15:54 * dajobe gets motion blur
15:16:29 <RSwick> color-coding nodes based on the W3C document status (WD, CR, ...)
15:18:01 <DanC> yeah... node label rules are pretty much the first thing you do with the circles-and-arrows tools I built.
15:20:07 <danbri> for pulling in properties into a table, could use brownsauce stuff?
15:20:11 * timbl wonders whether TVRamin could use an IRC speech stream in an earphone.
15:21:04 <dajobe> gosh, more gui fun - glyph factory
15:21:12 <dajobe> ZVTH it said on the title bar
15:21:40 <mattb> ZVTM is the zoomable user interface toolkit isaviz uses for rendering
15:23:28 <dajobe>http://www.w3.org/2001/11/IsaViz/
15:23:29 <dc_www2003> S: http://www.w3.org/2001/11/IsaViz/ from dajobe
15:23:35 <dajobe> S:|IsaViz home page
15:23:36 <dc_www2003> Titled item S.
15:23:57 <mattb> see http://www.w3.org/2001/11/IsaViz/graphstylesheets/gssmanual.html for gss and glyph stuff
15:24:26 <mattb> S:[http://www.w3.org/2001/11/IsaViz/graphstylesheets/gssmanual.html|gss and glyphs] in the manual
15:24:26 <dc_www2003> Added comment S1.
15:24:50 * dajobe enjoys working SVG for a change, not using Linux today
15:27:00 <RSwick> Emmanuel: IsaViz is an editor as well as a visualizer
15:27:07 <dajobe> alberto: are the slides for your talk earlier online?
15:27:11 <RSwick> [just like Amaya is an editor as well as a browser]
15:27:24 <alberto> yeah - just emailed to Eric
15:27:33 <dajobe> alberto: ok, great
15:27:38 <alberto> chumping URL here, hold on...
15:27:51 <dajobe> add to G:
15:28:51 <alberto> G:[http://www.asemantics.net/presos/www2003/|RDQL for real HTML slides] from this morning preso
15:28:51 <dc_www2003> Added comment G10.
15:32:53 <alberto> G6: demo3 - [http://demo.asemantics.com/biz/lmn/|Last Minute News (LMN)]
15:32:53 <dc_www2003> Replaced comment G6.
15:34:05 <libby> bye all
15:34:32 <mattb> and it's over
15:40:40 * mattb heads offline, cheers al.l
15:44:03 <dajobe> END OF CONFERENCE
15:44:06 <dajobe> ---
15:44:23 <dajobe> if you want to continue this, I suggest going to #rdfig or #foaf - we're friendly!
15:44:46 * zool smiles
15:45:06 <dajobe> ttfn
18:23:22 <ChanServ> [#www2003] This channel is publicly logged and blogged: http://www2003.xmlhack.com/
18:25:52 <ChanServ> [#www2003] This channel is publicly logged and blogged: http://www2003.xmlhack.com/
22:24:25 <edd> dc_www2003: die
22:24:33 <edd> </fin>
