Experimental IRC log swig-2009-06-02

Available formats: content-negotiated html turtle (see SIOC for the vocabulary)

Back to channel and daily index: content-negotiated html turtle

These logs are provided as an experiment in indexing discussions using IRCHub.py, Irc2RDF.hs, and SIOC.

00:29:44<Shepard>the audience didn't seem to get that reference though ;)
05:57:03<gromgull>gooooood morning #swig...
05:59:58<bengee>oh, wifi workie?
06:00:26<gromgull>kind of
06:00:34<gromgull>well - we're only just gettign started
06:00:40<gromgull>it will break when everyone connects I guess
08:05:02<mhausenblas>morning Web of Data
08:05:45<kwijibo>morning mhausenblas , still living it large ?
08:09:13<mhausenblas>kwijibo: all well here
08:09:37<mhausenblas>network is bit crap but seems to work
08:10:06<kwijibo>network crap - you must be at a european semantic web conference :D
08:11:08<mhausenblas>kwijibo: get lost :D
08:11:16<kwijibo>ouch
08:12:17<mhausenblas>SPOT09 was cool, tutorial went fine as well
08:13:02<mhausenblas>the SFSW talk where I presented Stephane's great Drupal stuff was also fine - good questions, comments
08:13:48<kwijibo>cool :) Drupal stuff = Neologism ?
08:14:16<gromgull>no - they use the drupal content... thingy to automagically provide rdfa
08:21:18<kwijibo>gromgull, mhausenblas anyone got a link ?
08:21:25<kwijibo>(to the drupal stuff)
08:22:09<gromgull>hmm - http://semanticscripting.org/SFSW2009/short_3.pdf
08:22:09<gromgull>is the paper
08:22:09<gromgull>perhaps there are links in there,
08:22:26<gromgull>my wifi is too damn slow to get it now though :)
08:22:26<gromgull>Actually - i might have a local copy
08:22:39<mhausenblas>kwijibo: see http://www.semanticscripting.org/SFSW2009/short_3.pdf
08:23:07<kwijibo>thanks gromgull + mhausenblas :)
08:26:44<gromgull>of course the lazy good-for-nothing organisers of sfsw didn't manage to get their proceedings on the conference usb stick
08:32:03<gromgull>besbes, are you in the demo session?
08:32:03<gromgull>by chance?
08:33:02<besbes>gromgull, sorry no, have to give my presentation in the applications session
08:33:43<mhausenblas>gromgull: do you know one of these SFSW09 OC slackers personally? :)
08:35:00<mhausenblas>ACTION sitting next to gromgull. better take care what I say here :)
08:35:09<gromgull>ah right
08:35:09<gromgull>No worries - i was going to point to you and mention gnowsis.com when talking about nepomuk shortly
08:35:34<besbes>gromgull, ah great -- but you still can do this :-) just refer to the guy with the baby ...
08:36:07<besbes>and, i will present gnowsis.com at the lightning talk session tomorrow
08:36:21<gromgull>ah nice
11:42:24<leobard>hello #eswc2009, any interesting news from crete? awaiting that you guys get good internet access...
11:45:32<mhausenblas>leobard: you can wait till the cows come home :D
11:49:19<bengee>http://personal.eswc2009.org/live
11:49:28<leobard>apparently dr haus has internet.... btw: the only pic on flickr from eswc sofar is this: http://www.flickr.com/photos/ciro/3584667532/
11:49:39<leobard>but I see bengee is twittering like a hamster: http://search.twitter.com/search?q=%23eswc2009
11:49:40<bengee>B:|chatter and pics from eswc2009
11:50:04<bengee>B:not (m)any pics yet, though..
11:53:31<hhalpin>I heard there's internet somewhere in california :)
11:54:05<hhalpin>in crete we're ordering emergency internet, but the pipes aren't big enough :)
11:56:11<Pipian>hhalpin: Where you at?
11:56:49<Pipian>ACTION is hacking on some linked data wonkery due to uninteresting talks this afternoon
11:57:47<mhausenblas>Pipian++ ... less talking, more executing ;)
12:12:02<Anchakor>tobyink: that uri ontology is a good idea, would some article/spec be made? really seems worth it... I like that it uses terms of the current web :)
12:13:41<tobyink>Anchakor: yes, it would probably be a good idea to create a more human-readable version, which could also explain some of the rationale behind it.
12:15:26<tobyink>e.g. rdf:values are literals. This is because { <x> owl:sameAs <y> ; u:identifier [ rdf:value <x> ] ; u:identifier [ rdf:value <y> ] . } doesn't provide much information when <x> and <y> have been smushed together by OWL reasoning.
12:24:21<hhalpin>tobyink: send me the URI ontology if you wish, put the URI of it here in SWIG.
12:24:34<hhalpin>tobyink: I've been working on a "ontology of resources" available here:
12:24:45<tobyink> http://purl.org/NET/uri
12:25:01<hhalpin>http://purl.org/NET/irw/
12:26:02<hhalpin>hmmm...maybe we should merge these.
12:26:11<hhalpin>I'll look at in depth at some point soon.
12:26:43<Anchakor>tobyink: there has been lot of talk about persistance of uris, I am wondering if w3c didn't do anything with some success in this direction...
12:26:46<hhalpin>ACTION is in lounge next to restaurant of Knossos hotel in HTML.
12:29:43<hhalpin>I believe the W3C has been reluctant to create its own PURL service.
12:29:59<hhalpin>but now with SemWeb taking off a bit, vocabulary hosting and persistence may be back on the agenda.
12:30:17<Anchakor>Im not talking about services, I am talking about ontology describing uris
12:30:28<hhalpin>Oh, for that there is:
12:30:37<hhalpin>http://www.w3.org/2006/gen/ont
12:30:42<hhalpin>From Tabulator
12:31:00<hhalpin>http://www.w3.org/TR/HTTP-in-RDF/
12:31:04<hhalpin>HTTP in RDF Working Draft
12:31:16<hhalpin>Not exactly what tobyink is talking about, but in same sort of space.
12:31:18<tobyink> http://purl.org/NET/irw/ - argh, Tabulator doesn't seem to like.
12:31:28<hhalpin>Can you give me error message?
12:31:40<hhalpin>It might be <!-- comment --> topbraid composer puts at end :(
12:31:47<hhalpin>Will try to fix shortly.
12:32:09<tobyink>It's the generic text/xml MIME type - it just doesn't realise it's looking at RDF at all.
12:33:49<Anchakor>I can see the ontology overlap
12:34:30<Pipian>ACTION is feeling quite lazy and is writing a PHP wrapper for roqet to make a SPARQL endpoint from a static Turtle file.
12:34:48<Anchakor>tobyink: would be good to map between the ontologies ;)
12:35:16<tobyink>Also, I have some HTTP-in-RDF extensions. http://buzzword.org.uk/rdf/http-ext#
12:35:55<tobyink>Anchakor: it's always good to map between ontologies. (Unless they assume fundamentally incompatible world views.)
12:40:48<hhalpin>Ah, OK, I'm not hosting the ontology, but will ping the host to do media type correctly.
12:52:40<mhausenblas>hhalpin: still waiting to receive a mail :)
13:26:44<melvster>is it reasonable to have an rdf policy of having no bnodes in documents?
13:27:18<kjetilkWork>melvster, yeah, I try to maintain that policy
13:27:21<Anchakor>I think it is
13:27:42<melvster>so lets say i have some bnodes, i need to zap them with rdf:ID or rdf:about, yes?
13:27:43<kjetilkWork>...in new things I write, I should add :-)
13:50:35<tobyink>melvster: there are certainly advantages to avoiding bnodes. e.g. a bnode-free graph can be reliably canonicalised.
13:51:46<kwijibo>talis platform doesn't have bnodes
13:57:02<iand>hard to link to a bnode
13:58:59<kwijibo>iand: hard but not impossible ?
13:59:22<iand>I guess I could think of a way using IFPs for indirection
13:59:24<reto>tobyink, canonicalizing with bnodes: you might now jcc's paper: http://www.hpl.hp.com/techreports/2003/HPL-2003-142.pdf
13:59:25<kwijibo>i guess you create your own bnode
13:59:39<iand>but that assumes that everyone agrees which properties are IFPs
13:59:48<kwijibo>then you could link somehow from that bnode back to the document containing the original document
14:00:07<reto>kwijibo, this would not be a bnode
14:00:25<reto>bnode do not have identifier, that's a serialization specific thing
14:00:34<kwijibo>no, i know
14:00:35<tobyink>reto: yes, I'm familiar with it. Some graphs with bnodes are canonicalisable, but *all* graphs without bnodes are canonicalisable.
14:00:37<reto>so it would be document+query
14:00:45<kwijibo>reto, sure
14:01:00<Pipian>Augh, come back SSH session
14:01:38<reto>kwijibo, afact all graphs with bnodes can be canonicaizaed, and for real world case in reasonable time
14:02:09<reto>jjc's algoritm is to optimize computation time, otherwise the meaningless triples would not be needed
14:05:49<kwijibo>i was thinking something along the lines of, if I see a bnode description of you in a <1.rdf> , and I want to link to it, I obviously can't because it doesn't have an identifier outside of its document scope, but I could make my own bnode like: : <#me> foaf:knows [ foaf:nick "Reto" ; foaf:page <1.rdf> . ]
14:08:30<Anchakor>bnode is sort of rdf wildcard
14:45:00<melvster>right im going to try and avoid bnodes completely then
14:45:15<melvster>any tips on what to replace them with? a good naming scheme?
14:46:36<Anchakor>I had some trouble deciding what to use as such "local" namespace too... how about http://bnodes/123 ? :)
14:58:19<Pipian_>http://www.telegraphis.net/data/countries/
14:58:42<Pipian_>F:|Linked Data URIs for countries by country code.
14:59:40<Pipian_>F:Along the lines of Geonames's http://geonames.org/countries/# URIs, except bearing semantics.
15:00:29<Pipian_>F:Also uses owl:sameAs to link together the sws.geonames.org URIs with dbpedia and the /countries/# URIs.
15:00:45<Pipian_>Probably should write up a VOID doc for it too.
15:00:53<Pipian_>In the same vein...
15:01:27<Shepard>Anchakor: but if everyone would use http://bnodes/ then that would totally defeat the point as you might clash with other people's URIs. unless you stick a UUID at the end of it or something :)
15:01:58<Pipian_>http://www.telegraphis.net/data/continents/ works for continents. I'm trying to clean up and make consistent URIs for capitals and subdivisions too, although there's still some mapping work to be done.
15:02:23<Pipian_>G:|Linked Data URIs for continents (by Geonames continent code).
15:02:58<Shepard>AFAIK http://geonames.org/countries/# is supposed to bear semantics and get used in other places in sws.geonames.org - it just needs fixing :)
15:03:16<Pipian_>I don't know why I got into making those URIs when airports was the way I got started with this.
15:04:18<Pipian_>Oh, it's linked in, but about all you can get out of it is (theoretically) querying for things in a country. You can't actually get the country data from it :)
15:04:35<Pipian_>At least last I checked there wasn't RDFa on that page ;)
15:05:44<Pipian_>Had to code up a few makeshift vocabs too, as there weren't any straightforward ones for some of the properties I wanted to represent (e.g. country codes)
15:06:01<Pipian_>SPARQL endpoints though, which Geonames doesn't have yet ;)
15:06:49<Pipian_>(A horrible kludge where I wrap a call to roqet on the command-line, but...)
15:07:02<kwijibo>ACTION wishes for a firefox add on that did colour syntax highlighting display of turtle/n3
15:07:10<Shepard>well, I think they wanted to make it so you can get data from there. but yeah, no SPARQL
15:07:31<kwijibo>or even just rendered it instead of making the file open in something else
15:07:44<Pipian_>D2R seemed a bit heavy-weight for a server where I started by running lighttpd :)
15:07:59<Shepard>hmm, the airport service on http://www.daml.org/2001/10/html/ doesn't seem to work anymore. but the airport ontology is still useful: http://www.daml.org/2001/10/html/airport-ont
15:08:10<Pipian_>Also, conneg on most of those URIs too.
15:08:30<Pipian_>(default text/n3, application/rdf+xml supported)
15:08:38<Pipian_>Yep.
15:08:42<PovAddict>ACTION wants such a firefox addon too, and a Kate plugin for syntax highlighting, and a standalone desktop app to see the graph sfructure of any rdf
15:08:56<Pipian_>That's why I've got http://www.telegraphis.net/data/airports/(airport-code)#(airport-code)
15:09:10<Pipian_>Which I'm trying to link back in to the countries, etc.
15:09:15<Pipian_>Thus the diversion.
15:10:58<kwijibo>why the slash and the hash ?
15:10:59<Pipian_>So, for example, http://www.telegraphis.net/data/airports/LGIR#LGIR
15:11:09<Shepard>it's a good addition to lingvoj. we need more linked data for all sorts of codes, especially ISO ones. I hope lingvoj will cover script codes as well at some point. and then there's the currency codes
15:11:32<Pipian_>Didn't want to have to mint two URIs by doing a redirect.
15:12:03<Pipian_>Yeah, I almost got caught up in doing currency codes... But I have to scrape them for each country, so that's going to be fun.
15:12:42<Shepard>would be nice if http://www.telegraphis.net/data/airports/ provided the complete list / dataset, if that's not too big :)
15:12:51<Shepard>some for continents and countries
15:13:29<Pipian_>Continents and countries do so. Airports I haven't gotten to yet, as I've not properly fixed up the links back to countries.
15:13:32<Shepard> http://en.wikipedia.org/wiki/ISO_4217 should be pretty complete
15:13:47<kwijibo>Pipian_:++ for the linked data (and for when you publish voiD about it ;) )
15:16:03<Anchakor>Shepard: than perhaps relative uri "._bnodes/123>
15:16:08<Pipian_>So yeah, I'm not sure if doing the # instead of a redirect is outdated these days.
15:16:18<Anchakor><._bnodes> I mean
15:16:21<Pipian_>it does seem like less work to resolve though.
15:16:48<Pipian_>(And still isolates the document URI from the object URI)
15:17:01<kwijibo>Pipian_: yeah, seems good
15:34:17<tobyink>Anchakor: URI fragments can contain slashes, so you could use <#bnodes/123> to represent a bnode.
15:36:56<tobyink>Pipian_: http://goddamn.co.uk/viewvc/misc/airports/
15:37:45<Pipian_>Hm, which dataset are they using?
15:37:52<Pipian_>I think it's the same if it's a CSV file.
15:38:09<Pipian_>Ah, they're using the broader one.
15:40:24<tobyink>they = me.
15:40:52<Pipian_>(I'm using http://openflights.org/data.html, and plan to use their Airlines one as well, whereas they're using one of the source datasets for that, OurAirports)
15:40:54<tobyink>It was just a quick mock-up for someone who wanted a demonstration of converting CSV data to RDF IIRC.
15:41:04<Pipian_>Aha
15:41:46<Pipian_>Anyway, gonna get going.
15:47:55<tobyink>phenny: The CSV file I used was ourairports.com/data/airports.csv
16:09:42<DanC>ACTION wishes for a UML diagram of this PDB stuff
16:17:34<mhausenblas>DanC: PDB?
16:17:41<DanC>.g PDB protein
16:17:41<phenny>DanC: http://www.rcsb.org/
16:17:53<mhausenblas>ACTION notes that DanC always has new cool stuff, so eager to learn
16:18:12<mhausenblas>ta
20:07:05<Anchakor>phenny: tell tobyink I was wondering if rather not use from sub-delims which is part of pchar, as it would be more semanticly appropriate maybe... ( http://tools.ietf.org/html/rfc3986#section-3.5 ) / or . is not much difference
20:07:05<phenny>Anchakor: I'll pass that on when tobyink is around.
21:13:19<drobilla>is it possible to express, in owl, an instance of whatever class must be the /object/ of such and such properties on some other object?
21:13:58<drobilla>scenario is: the class is a group of things that do not have URIs, and those things have 'roles' in the group
21:14:10<drobilla>to be valid, a group must have such and such set of roles
21:14:46<drobilla>(yes, this is weird, but the things do not and can not have URIs for other reasons)
23:03:59<gsnedders>Is there any way to use one SPARQL query as the FROM of another? The grammar doesn't allow it, so I'm guessing I'd need to do processing in-app to create the FROM statement.

Back to channel and daily index: content-negotiated html turtle