Experimental IRC log swig-2009-05-21

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.

05:22:28<mhausenblas>morning Web of Data
05:28:06<mhausenblas>kennyluck around?
05:28:41<mhausenblas>ACTION saw your http://chatlogs.planetrdf.com/swig/2009-05-20.html#T23-29-51 query
05:30:09<mhausenblas>we just recently did that, see http://sw-app.org/pub/seke09-ld2sd.pdf - looks like we used BAETLE for it
05:31:31<mhausenblas>actually still improving the base of software artefacts that we support and the interlinking as well (which is quite interesting, btw, e.g. how to link SVN with an issue tracker, src code, etc.)
05:32:15<mhausenblas>you may wanna ping Aftab re the details, he's mainly working on it ...
05:32:51<mhausenblas>phenny, tell kennyluck res SVN logging see http://chatlogs.planetrdf.com/swig/2009-05-21.html#T05-28-18
05:32:51<phenny>mhausenblas: I'll pass that on when kennyluck is around.
07:08:00<mhausenblas>http://webofdata.wordpress.com/2009/05/21/discussing-powder/
07:08:10<mhausenblas>A:| Discussing POWDER and discovery mechanisms on the Web …
08:53:58<tobyink>kennyluck: RDFizing a SVN log - just use RSS 1.0. Each commit is an rss:item, the revision number is the rss:title, the log message is its rss:description, the date is its dc:date, the dc:creator is the username of the person that checked the commit in, and the rss:link is a link to a subversion web interface (preferably a link to that specific revision).
08:54:49<mhausenblas>not bad, toby
08:55:10<mhausenblas>ACTION wondering if we can do more complex stuff such as branches, etc with this as well
08:55:11<kennyluck>yes, I am in fact using dc:creator and dc:date
08:55:12<phenny>kennyluck: 05:32Z <mhausenblas> tell kennyluck res SVN logging see http://chatlogs.planetrdf.com/swig/2009-05-21.html#T05-28-18
08:55:25<kennyluck>Not sure about what type to use.
08:55:35<mhausenblas>anyway I like the idea. it's so simple that it really makes sense
08:55:39<kennyluck>You think rss:item is a good one? tobyink?
08:55:52<kwijibo>mhausenblas: voiD call in 5 mins ?
08:55:57<mhausenblas>yep
08:56:06<mhausenblas>good morning kwijibo, btw ;)
08:56:21<mhausenblas>though cygri in not yet around, slacker
08:56:51<mhausenblas>anyway, if he doesn't show up I gonna call him on his mobile :)
08:57:37<mhausenblas>ACTION off to grab a coffee before the call
08:58:29<tobyink>kennyluck: there are probably people who would say that the rss:item and the SVN commit are "separate things" and that in fact, the rss:item and commit should just be linked by a foaf:primaryTopic.
08:59:10<tobyink>But I disagree - I think the definition of rss:item is vague enough that it's possible for a SVN commit to be both a commit *and* and rss:item.
08:59:25<tobyink>Much as I am both a foaf:Person and an owl:Thing.
08:59:40<tobyink>Because everything is an owl:Thing.
08:59:43<kennyluck>sure, by the way a rdfs:label instead of rss:title
09:00:09<kennyluck>I don't quite see the difference between rdfs:label foaf:name dc:title and so on
09:00:26<tobyink>The advantage of using rss:item, rss:title, rss:link is that you get support in feed readers as a bonus
09:00:40<kwijibo>kennyluck: there isn't one in the semantics
09:00:40<kennyluck>I see, that makes sense
09:00:50<kwijibo>as far as i've been able to tell
09:01:07<kwijibo>there are loads of labelling properties that all mean the same thing
09:01:16<kwijibo>i prefer rdfs:label
09:01:27<kennyluck>I kind of think that foaf:name is a owl:FunctionalProperty while rdfs:label is not
09:01:35<kennyluck>but this is not in the FOAF spec
09:01:54<kwijibo>nope
09:02:06<tobyink>kennyluck: I don't think so. <#i> foaf:name "Toby Inkster", "Toby A Inkster", "Toby Andrew Inkster".
09:02:09<kwijibo>i dunno, i think you can sensibly have more than one foaf:name
09:02:35<tobyink><#bw> foaf:name "Bruce Wayne", "Batman" .
09:02:36<kennyluck>I don't know, I think foaf:name should be your passport name or something standard.
09:02:39<kennyluck>tobyink
09:02:53<kennyluck>or you can just use rdfs:label.
09:03:02<tobyink>I revealed his secret. Sorry.
09:03:14<kennyluck>Ha-Ha
09:04:42<tobyink>Yay! My rebuild of my railway data last night fixed some geocoding issues. It was getting facts mixed up about Euston, St Pancras and King's Cross stations because they're all within a few hundred metres of each other.
09:04:51<kennyluck>tobyink, your idea about making the SVN logs readable to a RSS reader makes a lot of sense, thank you!
09:05:26<tobyink>No problem. I have been thinking a lot recently how virtually everything interesting should have rdf:type rss:item.
09:06:08<kennyluck>Yep, agreed, RDF is to augment interoperability.
09:06:49<libby_>foaf:name is not a functional prop, no, it's nowhere near unique enough
09:07:38<tobyink>libby: uniqueness would make it an *inverse* functional property, no?
09:08:06<libby>yeah soz
09:08:09<libby>ACTION not awake yet
09:08:16<libby>despite it being after 10
09:08:33<tobyink>ACTION is interested in this VoCamp libby's putting on in September.
09:08:37<kennyluck>the point is, I think, there has to be a special character about foaf:name to tell it apart from rdfs:label, libby.
09:08:51<libby>why?
09:09:08<kennyluck>Or we just end up with lots of predicates with no interoperability.
09:09:35<mischat>ACTION wants to come to vocamp too
09:09:37<libby>yeah, vocamp should be cool - I'm not super involved right now but will be (along with shellac, tommyh etc)
09:09:46<tobyink>Pretty much every vocab makes the mistake of duplicating some basic stuff which already exists in RDFS or Dublin Core.
09:10:10<libby>I wouldn't worry too much
09:10:22<kennyluck>yes, and I don't like it. That's why I like tobyink's idea.
09:10:34<libby>ACTION believes there's a whole lot of fussing about vocabs and people should jfdt
09:10:55<libby>jdfi
09:10:56<libby>sigh
09:11:05<libby>ACTION is just trying out my3G dongle really
09:11:12<tobyink>In a world with RDFS and OWL reasoning, it's not a big issue. But if you're not using them, then having hundreds of different predicates to check to just find a label for something can be annoying.
09:12:11<libby>just hardcode in a couple of the main ones and if your app is intersting enough people will converge on them
09:12:23<kennyluck>In the current world, I think vocab makers just forgot to make rdfs:subPropertyOf links to all other properties.
09:28:57<kennyluck>mhausenblas, the Baetle vocab looks like something I wanted to find, thank you.
09:29:53<kennyluck>I have a question. Once I make these triples as Linked Data. Am I going to be able to use the application developed in your paper?
09:36:33<tobyink>kennyluck: I think mhausenblas said he was on a conference call. You might want to get phenny to ask him.
09:36:45<kennyluck>Thanks
09:37:59<kennyluck>By the way, tobyink, my last concern would be, does a RSS reader actually parses RDF or it just find XML tags. Since my logs have extra triples in it, I wonder it will work out or not.
09:38:49<tobyink>Mostly they do the tag soup thing. Are you using an RDF serialisation library to output your RDF/XML, or outputting it by just printing strings?
09:40:07<kennyluck>RDF serialization library, tobyink.
09:43:58<tobyink>Ah, which library?
09:44:07<kennyluck>Tabulator
09:44:37<tobyink>Don't know much about Tabulator's internals, so don't know how much of this is do-able.
09:45:00<kennyluck>Hmm... I'll try to figure it out.
09:45:22<tobyink>But the general tricks is to make sure that the RSS 1,0 namespace is the *default* namespace, so that you can just use <link>, <title>, etc.
09:45:47<tobyink>And make sure that dublin core is bound to "dc" and not something exotic.
09:47:45<tobyink>Oh, and make sure that you actually use <item> rather than <rdf:Description><rdf:type rdf:resource="...item">.
09:48:18<kennyluck>Alright, that sounds tough though.
09:50:31<tobyink>As I said, I don't really know the internals of Tabulator. I hacked ARC2 to follow these rules for libre.fm's feeds and it wasn't especially difficult - but then again, ARC2's RDF/XML output was quite pretty to begin with.
09:53:39<tobyink>melvster_: This looks pretty cool <http://www.w3.org/TR/2008/WD-geolocation-API-20081222/> - Opera have recently released a preview implementation of it.
09:55:01<tobyink>Possible foaf.me use - query the user's geolocation, and add this to their foaf file: <#me> foaf:based_near [ geo:lat 12.345 ; geo:long 123.45 ] .
09:57:19<Anchakor>kennyluck: I hope you intend to add a bit of semantics - not just use rss, so I can search for all logs of some topic... :)
09:58:09<kennyluck>Yes, Anchakor, that's a must.
09:58:42<kennyluck>For now there are triples that "Revision 12345" --affects--> <http://some.document>
09:58:52<tobyink>Another thing you could include is a link from the rss:item to each file changed. I'm not entirely sure of the best ...
09:58:53<Anchakor>was, just making sure... I would probably define my own vocabulary and export some rss
09:59:00<tobyink>You beat me to it kennyluck!
09:59:06<kennyluck>so you can find all the revisions for a document
09:59:27<kennyluck>This is the current file, FYI. http://dig.csail.mit.edu/svn/recent
09:59:57<kennyluck>I'll try to make it a RSS compliant feed, while it's now in N3
10:02:20<tobyink>dc:relation is a sufficiently vague property that it could be used to link from the <item> to the files changed. You may wish to define your own term which is an rdfs:subPropertyOf that. (http://open.vocab.org/) is a good place for defining quick one-off terms.
10:04:09<kennyluck>tobyink, I am trying to find the exact predicate in Baetle, the vocab michael mentioned.
10:04:32<kennyluck>I don't feel like creating any new terms. I don't like it.
10:06:04<tobyink>I've taken a look at Baetle, and it looks interesting, but mostly related to bug/issue tracking rather than source code repository logging.
10:06:46<tobyink>Someone should look at DOAP, Baetle and repository logging under a bigger umbrella.
10:06:49<kennyluck>ACTION is looking for it in http://baetle.googlecode.com/svn/ns/Baetle.ewl.n3
10:07:13<kennyluck>Well, we should have a DOAP community channel or we can just discuss it here
10:07:36<kennyluck>It's nice to use predicates from difference namespace. That's the spirit of RDF
10:08:30<libby>yeah I sometimes think it leads to bad things though
10:08:37<libby>like semantic drift
10:09:00<libby>and huuuuge namespace declaartions
10:09:16<tobyink>Everything in moderation.
10:23:41<danbri>DOAP is very welcome here, kennyluck
10:24:17<danbri>re ns proliferation ... i just wrote big long mail on that
10:24:37<danbri>http://lists.w3.org/Archives/Public/www-archive/2009May/0051.html
10:24:50<danbri>B:|discussion of foaf and overlaps, re the foaf "vs" vcard thing
10:28:27<Anchakor>huge namespace declarations shouldn't be a problem or whole rdf is flawed
10:29:28<tobyink>I don't really follow www-archive because it seems like a very odd mailing list. But interesting things do tend to come up on, so perhaps I should be following it.
10:30:13<danbri>in some contexts they're a problem, in some they're not
10:30:15<danbri>that's quite natural and no cause for melodrama in either direction
10:30:27<danbri>there are problems with everything, and yet somehow the world doesn't blue-screen
10:32:52<tobyink>B: Talks about RSS 1.0 being designed too small. Taking modularisation too far. (Case in point: you need to use dc:date.)
10:38:59<kennyluck>danbri, do you mind to spend some time on incorporating all the FOAF translations you got in http://svn.foaf-project.org/foaftown/foaf18n/ to the RDF/XML you got when fetching http://xmlns.com/foaf/spec/ .
10:39:35<kennyluck>Or maybe you would allow me to do it? If you are busy working on something else.
10:50:11<kennyluck>ACTION is heading out
10:50:12<danbri>kenny, they'll be in the next version!
10:50:16<danbri>the delay was mostly fixing the spec-management scripts (plus getting hacked)
10:50:24<kennyluck>danbri, cool.
10:50:59<tobyink>danbri: did you see my messsage in #swxg?
11:14:11<kennyluck>tobyink, just curious, have you used your tool (swigtion?) to scrape web sites?
11:14:25<kennyluck>in to RDF, I mean.
11:32:20<tobyink>Yes, on occasion.
13:12:39<idmclean>phenny tell tobyink What do we need to define next in the ontology, and what do we need to re-examine or further define?
13:13:12<idmclean>phenny, tell tobyink What do we need to define next in the ontology, and what do we need to re-examine or further define?
13:13:12<phenny>idmclean: I'll pass that on when tobyink is around.
13:26:46<connolly_>what's the XSLT magic for omitting certain namespace declarations from the output?
13:33:21<Shepard>DanC: exclude-result-prefixes="<space-separated list of prefixes>" on the xslt:stylesheet element
13:33:35<DanC>right; thanks. found it
14:13:19<mhausenblas>hu? IBM has patented the write-Web of Data?
14:13:21<mhausenblas>http://www.freepatentsonline.com/7519599.html
14:13:30<mhausenblas>C:| System and method for scalable distribution of semantic web updates
14:14:11<mhausenblas>C: I wonder how this might effect our work on an open write-enabled Web of Data
14:16:12<dajobe>please don't mention patents - software people should not look
14:17:03<mhausenblas>dajobe: sorry?
14:17:16<mhausenblas>what do mean? that I don't have to worry or ...
14:17:24<mhausenblas>ACTION *is* worried
14:17:37<kwijibo>that's mental
14:17:44<mhausenblas>hehe
14:18:02<mhausenblas>seriously, dajobe, are you suggesting I should ignore that, yes?
14:18:10<dajobe>yes
14:18:16<mhausenblas>hm
14:18:18<kwijibo>because it doesn't matter?
14:18:21<dajobe>and never go looking for so-called software parents, never read them
14:18:23<kwijibo>or as a boycott of patents ?
14:18:40<dajobe>because it can be knowing infringment if they ever sued you
14:18:42<mhausenblas>I didn't look for them but stumbled upon them
14:18:55<mhausenblas>ah. I get it.
14:19:06<kwijibo>ah
14:19:21<dajobe>also, but slightly unrelated, software patents are a mega dumb idea, and you should ignore them for that reason too
14:19:40<dajobe>ACTION member FSF & EFF
14:20:25<kwijibo>dajobe: only software patents ?
14:21:20<dajobe>no, but that's probably off topic for here.
14:21:43<dajobe>could somebody patent data? don't think so
14:22:02<dajobe>ACTION heads back to Where 2.0 ...
14:22:17<mhausenblas>ta, dajobe. saved my ass ;)
14:22:39<Anchakor>ACTION is lucky, he didn't look :)
14:23:58<mhausenblas>C:= http://www.endoftheinternet.com/
14:24:23<mhausenblas>C:| End of the Internet
14:25:08<mhausenblas>C1 = in case you're overworked
14:25:22<mhausenblas>C1: in case you're overworked
14:25:39<Anchakor>someone should secretely clean the logs of this and replace it with usual connect/disconnect stuff ;)
14:26:41<mhausenblas>ACTION thinks he just did - at least one part ;)
14:38:59<kwijibo>mhausenblas: are you aware of anyone harvesting voiD descriptions ?
14:39:50<kwijibo>sindice has the ones frm rkbexplorer, and one of mine, but not all - even though I'm sure i've told it about others
14:40:32<kwijibo>and you wrote a paper about a voiD search engine right ?
14:41:31<mhausenblas>ahm, not to my knowledge, no (re harvesting)
14:42:07<mhausenblas>I dunno if it makes sense, as basically every indexer that indexes RDF automatically as a voiD harvester ;)
14:42:36<tobyink>FWIW, it's possible to tell logger to stop logging ("logger, stop") and start up again ("logger, start").
14:42:36<phenny>tobyink: 13:13Z <idmclean> tell tobyink What do we need to define next in the ontology, and what do we need to re-examine or further define?
14:42:38<mhausenblas>paper? nah, not really, where? only the one article in NodMag
14:42:45<kwijibo>well - it makes sense to aggregate and provide a query service surely ?
14:43:04<mhausenblas>too late for me tobyink I guess ;)
14:43:33<tobyink>Though of course, many IRC clients perform logging, so dajobe's public logs are not likely to be the only record of this channel.
14:43:55<mhausenblas>yeah, sure, kwijibo, that would make sense, if you provide a special service (but maybe within existing infrastructure, not as standalone service)
14:44:04<kwijibo>mhausenblas: what about ding ?
14:44:14<mhausenblas>that's a ranking algorithm
14:44:21<mhausenblas>based on voiD
14:44:38<mhausenblas>no, what we'd need is something along the line of LDE
14:44:49<mhausenblas>.g LDE voiD explorer
14:44:50<phenny>mhausenblas: http://semanticweb.org/wiki/VoiD
14:44:53<mhausenblas>no
14:45:08<mhausenblas>see http://ld2sd.deri.org/lde/
14:45:11<kwijibo>g LDE voiD explorer deri
14:45:29<kwijibo>why the asterisk in the title ?
14:45:44<mhausenblas>and rather than having a static base it would query semantic indexers for it
14:45:57<mhausenblas>kwijibo: just to highlight what LDE stands for ;)
14:46:36<mhausenblas>so, I'm gonna look into it (coverage of voiD indexing by semantic indexers such as Falcon-S, Watson, SWSE, and Sindice)
14:46:44<mhausenblas>see what's possible, there
14:46:55<mhausenblas>and rewrite the interface of LDE to be fed by them, ok?
14:47:08<kwijibo>ptsw ought to index them too
14:47:21<kwijibo>mhausenblas: lde is pretty slick
14:47:23<mhausenblas>yep, good point
14:47:39<kwijibo>mhausenblas: but it only lists one - and I know I've pnged it with more than that
14:47:40<mhausenblas>slick in the sense of?
14:47:47<kwijibo>nice
14:47:59<mhausenblas>nah. plain. simple. stupid, yes ;)
14:48:14<mhausenblas>it misses a lot. damn, if I only had more time ;)
14:48:40<Anchakor>?? LDE
14:48:49<mhausenblas>take for example the visualise tab. played around with misusing the Google API for that ;)
14:49:03<mhausenblas>Anchakor: a linked dataset explorer based on voiD descriptions
14:49:48<mhausenblas>see http://ld2sd.deri.org/lde/ - but don't expect too much :)
14:49:54<kwijibo>mhausenblas: http://api.sindice.com/v2/search?page=1&q=*+%3Chttp%3A%2F%2Fwww.w3.org%2F1999%2F02%2F22-rdf-syntax-ns%23type%3E+%3Chttp%3A%2F%2Frdfs.org%2Fns%2Fvoid%23Dataset%3E&qt=advanced
14:50:31<kwijibo>some of the uris are wrong though, because rkbexplorer changed their URI pattern
14:50:47<kwijibo>so if it ends in /id it shoudl really end in /id/void ( think)
14:52:01<Anchakor>mhausenblas: the visualization somehow doesn't work :)
14:53:08<mhausenblas>Anchakor: yep. see above ... if I only had more time ...
14:53:19<mhausenblas>ACTION BRB ... grab some coffee
15:08:53<mhausenblas>OpenLink chaps? seems http://lod.openlinksw.com/void/ is down ...
15:09:29<kidehen>mhausenblas: yes it is, upgrading the binary in relation to the SPARQL DESCRIBE issue etc..
15:09:43<kidehen>mhausenblas: will be back online in a few minutes
15:10:05<kidehen>mhausenblas: this is the issue affecting the /sparql endpoint
15:10:35<mhausenblas>thanks, kidehen
15:10:41<kidehen>mhausenblas: eta about 15 - 20 mins
15:10:50<mhausenblas>great, that's cool
15:11:07<kwijibo>ACTION wonders if RDF crawlers/browsers send referrer headers
15:11:29<mhausenblas>ACTION gonna switch over to use OL voiD graph in LDE and in a second step to query semantic indexers to get live data
15:11:42<kwijibo>ACTION also wonders about sending somehow, the type of link that was followed
15:12:12<mhausenblas>would be cool if all semantic indexers had an Atom feed that would inform me about new triples of form * rdf:type void:Dataset
15:12:33<mhausenblas>kwijibo: ? sorry - was that for me?
15:12:36<kwijibo>mhausenblas: how many are there ?
15:12:50<kwijibo>mhausenblas: just general pondering
15:13:17<mhausenblas>min. 6?
15:13:20<mhausenblas>see http://esw.w3.org/topic/TaskForces/CommunityProjects/LinkingOpenData/SemanticWebSearchEngines
15:13:24<kwijibo>mhausenblas: I remember Cygri wanted to add a VoidDescriptionDocument class
15:13:43<mhausenblas>yep
15:14:51<kwijibo>and ldodds told me yesterday that he thought typing the object uri was a good way of getting round the vagueness of rdfs:seeAlso links
15:15:38<mhausenblas>ah
15:15:39<kwijibo>so maybe a better pattern to <> terms:partOf <void.n3>
15:15:49<mhausenblas>mhm
15:15:55<kwijibo>i mean
15:16:04<kwijibo> <> terms:partOf <void.n3#Dataset>
15:16:07<kwijibo>could be
15:16:25<mhausenblas>we could write a little service that polls the major SI and provides an Atom feed
15:16:40<mhausenblas>for the time till they all get it ;)
15:16:48<kwijibo><> rdfs:seeAlso <void.n3> . <void.n3> a void:VoidDocument
15:17:09<kwijibo>mhausenblas: I think morph.talis.com might be able to do this without any coding
15:17:13<kwijibo>let me see
15:17:20<kwijibo>the feed i mean
15:17:37<kwijibo>because it lets you merge graphs and output as RSS
15:17:41<mhausenblas>that would be awesome
15:17:52<mhausenblas>give it a try, please
15:17:52<kwijibo>and ARC understands atom I think
15:18:01<mhausenblas>mmmm, guess so
15:18:16<kwijibo>or maybe just rss 2
15:18:18<mhausenblas>I'd also buy RSS FWIW ;)
15:18:39<mhausenblas>but how do you poll?
15:19:14<kwijibo>mhausenblas: how do you mean ?
15:19:25<kwijibo>I guess your rss reader does the polling, no ?
15:20:44<kwijibo>fgiasson: is there any way to get a list of urls with void:Datsaet classes in them from PTSW ?
15:21:41<kidehen>mhausenblas: lod.openlinksw.com is back online
15:22:12<mhausenblas>ta kidehen
15:22:37<mhausenblas>kwijibo: no, I meant polling the semantic indexer for it ...
15:23:20<tobyink>kwijibo: For Swignition I seem to remember being very careful to get referer headers right. Sending referer headers enables some pretty awesome applications of GRDDL.
15:23:52<kwijibo>mhausenblas: your rss reader requests rss from morph, which requests data from the sem indexes and merges the graphs into an rss feed
15:24:11<fgiasson>kwijibo: I could manually, but it appears that only one document with the Void namespace has been indexed by PTSW atm
15:24:12<kwijibo>your feed reader then displays any new ones i guess
15:24:39<kwijibo>fgiasson: that's odd - I think it used to have at least 2
15:24:45<tobyink>RSS 1.0 needs a marketing refresh so that people can rediscover its awesomeness. I vote we start calling it RSS5.
15:24:54<kwijibo>and i've pinged it separately with problaly at least 3
15:24:58<kwijibo>:)
15:25:04<kwijibo>tobyink++
15:25:18<fgiasson>kwijibo: don't know; would have to test with a couple of URIs
15:25:42<mhausenblas>kwijibo: got it now :)
15:25:53<tobyink>Namespace URI stays at 1.0 though, because as with other *5 standards we care about backwards-compatibility.
15:26:02<mhausenblas>FWIW, Falcon-S and Watson seem to have none
15:27:51<dajobe>tobyink: I suggest RSS∞ so Winer can't just add 1 again
15:30:10<Anchakor>why is rss so awesome? is it more awesome than rdf?
15:30:27<tobyink>RSS 1.0 is RDF.
15:30:33<mhausenblas>Anchakor: the workflow is awesome
15:30:52<mhausenblas>I can use it as API and for manual inspection, if I like
15:31:02<tobyink>dajobe: Well, Winer could still add 1 to ∞, it just wouldn't make any difference.
15:31:37<mhausenblas>of course one could also publish it in AtomOwl ;)
15:31:39<kwijibo>me breaks swse
15:31:40<kwijibo>http%3A%2F%2Frdfs.org%2Fns%2Fvoid%23Dataset
15:31:42<Anchakor>tobyink: why not then make RSS5 and use turtle as serialization?
15:31:46<kwijibo>http://swse.deri.org/detail?focus=void%3ADataset
15:32:08<kwijibo>what letter was that ?
15:32:15<mhausenblas>uhuh, gotta tell Andreas ;)
15:32:25<mhausenblas>infinity?
15:32:45<PovAddict>D
15:33:15<tobyink>Infinity, yes.
15:33:15<kwijibo>D: test uri for swse
15:33:31<kwijibo>ta PovAddict
15:33:43<kwijibo>ACTION has smilies turned on :p
15:33:45<tobyink>Anchakor: because that would destroy compatibility with real-world feed readers.
15:34:07<kidehen>kwijibo: you can retest your BBC query against lod.openlinksw.com/sparql
15:34:15<kwijibo>ta kidehen
15:34:21<tobyink>kwijibo: infinity appears as a smile? What kind of weirdly disfigured mouth is that based upon?
15:34:45<kwijibo>no, D: appears as a sad face
15:35:04<kwijibo>i wanted to chump the url
15:35:11<kwijibo>and it gave me a D
15:35:17<Anchakor>tobyink: I know, but my original question wasn't really answered anyway - I should reform it: why and how should rdf linked data system use rss?
15:35:36<mhausenblas>D:| test URI for SWSE
15:35:43<mhausenblas>like that?
15:35:53<kwijibo>sorry, thanks mhausenblas
15:36:04<mhausenblas>D1: kwijibo's breaking the system
15:36:08<mhausenblas>:P
15:37:23<tobyink>Anchakor: a linked data publisher can use RSS 1.0 in RDF/XML to be able to provide data which is compatible with RDF tools and feed readers simultaneously. (Similar to how they can use RDFa to provide data which is compatible with browsers).
15:42:17<PovAddict>D:|test uri for swse
15:42:20<PovAddict>D1:""
15:42:26<PovAddict>oh wtf
15:42:29<PovAddict>someone had already done that
15:42:30<PovAddict><.<
15:42:49<PovAddict>sorry :p
15:43:06<PovAddict>D:| test URI for SWSE
15:43:06<PovAddict>D1: kwijibo's breaking the system
15:43:25<PovAddict>D: kwijibo's breaking the system
15:45:03<kwijibo>ACTION wonders why rss 1.0 was developed as an xml serialisation, and not just an RDF vocabulary ?
15:48:42<kwijibo>danbri: ? iand ldodds ?
15:49:38<ldodds>to try and hit a sweet spot of working in both the XML and RDF worlds
15:50:03<PovAddict>"RSS 0.90 was the original Netscape RSS version. This RSS was called RDF Site Summary, but was based on an early working draft of the RDF standard, and was not compatible with the final RDF Recommendation."
15:50:09<ldodds>and to build on previous work, which was already an XML serialization
15:51:09<PovAddict>rss 2.0 sux
15:51:52<PovAddict>if I want an XML format for feeds, not using RDF, I'd go for Atom...
15:53:04<kwijibo>ACTION wonders what the point of rss 2.0 and atom syndication were
15:53:35<kwijibo>and if/why site owners should bother providing them
15:53:49<tobyink>RSS 2 is essentially Dave Winer's continuation of RSS 0.9x but under a 2.0 name for marketing purposes.
15:54:06<kwijibo>i've never noticed a difference whichever I choose to subscribe to
15:54:30<tobyink>Atom was born out of frustration with the many incompatible flavours of RSS.
15:55:11<kwijibo>it's sort of like every time i went to a webpage, it asked me - do you want html 4, xhtml 1, or xhtml 1.1 ?
15:55:20<tobyink>Essentially most RSS readers just suck anything they get pointed at (often including Atom) as tag soup.
15:56:08<PovAddict>wtf
15:56:10<PovAddict>not even a real XML parser?
15:56:47<tobyink>Some will use a real XML parser, but even that's a risky choice with some of the crummy feeds in the wild.
15:57:17<PovAddict>if I wrote a feed reader I wouldn't care about that :D
15:57:25<tobyink>As soon as one implementation parses a broken feed, "standards compliance" becomes a race to the bottom.
15:57:36<dajobe>I did write a feed parser for raptor, it's name 'rss-tag-soup'
15:57:47<kwijibo>haha :)
15:57:48<tobyink>And it's pretty awesom.
15:57:56<gsnedders>This is why quite a lot of us have been arguing there should be some standardized non-fatal error handling for XML
15:57:57<PovAddict>similarly, I wouldn't change reader because it can't parse a broken feed
15:58:17<gsnedders>The race to-the-bottom happened long ago, we're just living in a world of undefined behaviour.
15:58:18<kwijibo>i like dajobe's tagline for triplrs rss feature "good rss from bad"
15:58:31<tobyink>dajobe: though if I were you I'd change the Atom parsing to use bblfish's AtomOwl vocab.
15:58:40<PovAddict>gsnedders: putting invalid code into a tag soup reader is undefined behavior
15:58:45<dajobe>tobyink: I'm in hte middle of a rewrite of the parser
15:58:50<PovAddict>gsnedders: putting invalid code into a XML parser is defined: fatal error
15:58:52<dajobe>but I'm not likely to use the atom owl format
15:58:54<dajobe>vocab
15:59:17<gsnedders>PovAddict: Sure. it's defined for XML parsers. That doesn't have much relevance for feed parsers.
15:59:31<PovAddict>why not, feeds are XML
15:59:37<dajobe>tobyink: you can see the tests I'm working to at present at http://svn.librdf.org/view/raptor/trunk/tests/feeds/
15:59:38<tobyink>Well, at least stop using the http://www.w3.org/2005/Atom namespace because it results in weird looking URIs.
15:59:41<gsnedders>PovAddict: In the real world, they aren't.
15:59:47<PovAddict>too bad
16:00:00<dajobe>tobyink: wierd looking is not a bug
16:00:12<gsnedders>PovAddict: Oh well, have fun with the small number of feeds that work in a fully conforming implementation.
16:01:01<kwijibo>mhausenblas: have you figured out how to get datasets form falcon ?
16:01:06<kwijibo>*from falcon
16:01:10<mhausenblas>kwijibo: no :(
16:01:11<tobyink>dajobe: is "using URIs that were never intended as rdf:Properties as rdf:Properties" a bug?
16:02:06<kwijibo>what is the difference between an object and a concept ?
16:02:17<dajobe>tobyink: hard to say what they intended. it was all defined as xml, so how to form rdf predicate/class uris wasn't on their mind. that they chose a namespace not ending in # or / was probably either dont-care or just being awkward
16:02:24<dajobe>ACTION concentrates on wher2.0
16:02:49<gsnedders>danbri: Nice reply to Doug on -archive :)
16:02:49<mhausenblas>kwijibo: schema - instance level, I'd say
16:03:07<PovAddict>nowhere says a XML namespace URI should end in a # or /
16:03:17<kwijibo>mhausenblas: falcons doesn't seem to have indexed void
16:04:04<mhausenblas>kwijibo: correct. see http://chatlogs.planetrdf.com/swig/2009-05-21.html#T15-26-14
16:04:28<kwijibo>mhausenblas: i mean, not even the vocab
16:04:32<bblfish>that's why I spent a lot of energy on AtomOwl
16:05:07<bblfish>still there: http://bblfish.net/work/atom-owl/2006-06-06/AtomOwl.html
16:05:55<mhausenblas>kwijibo: yep, also stats seem to indicate they're a bit behind time ;)
16:06:26<kwijibo>mhausenblas: stats ?
16:07:35<mhausenblas>kwijibo: see http://iws.seu.edu.cn/services/falcons/statistics.jsp
16:07:51<kwijibo>mhausenblas: ah
16:08:08<kwijibo>mhausenblas: did you also see that dataincubator projects are using voiD ?
16:08:23<kwijibo>eg: http://oecd.dataincubator.org/
16:08:27<mhausenblas>yep, great!
16:08:42<mhausenblas>I think I've twitted about it some time ago
16:09:06<kwijibo>ah really? I only really follow identica lately
16:09:16<kwijibo>this one is mine
16:09:18<kwijibo>http://climb.dataincubator.org/.html
16:10:17<mhausenblas>yep, same with this ;)
16:14:56<kwijibo>mhausenblas: can't find your tweet - how old ?
16:15:55<mhausenblas>lemme see ...
16:19:37<mhausenblas>kwijibo: hm, seems I must have erred - maybe on swig?
16:21:18<mhausenblas>ah, yeah, http://chatlogs.planetrdf.com/swig/2009-04-24.html#T06-59-49
16:21:29<kwijibo>http://chatlogs.planetrdf.com/swig/2009-04-24.html#T06-59-49
16:21:34<kwijibo>:)
16:23:04<kwijibo>mhausenblas: if I use rdfs:seeAlso instead of terms:partOf, will that make my dataset resource discoverable to tabulator ?
16:25:51<mhausenblas>kwijibo: good question. oh, oshani just left ...
16:26:11<mhausenblas>ACTION BRB
17:30:06<kidehen>http://tr.im/m0PT
17:30:36<kidehen>G: github project with all the xslt's used in our sponger cartridges
17:31:24<kidehen>G: naturally these should also usable by GRDDL processors
17:38:00<Anchakor>ACTION wonders what github means as "social coding"
17:38:39<_psychic_>you can fork other projects and easily share code
17:38:42<PovAddict>Anchakor: taking advantage of the 'social' hype
17:38:48<PovAddict>by using that word in its marketing
17:39:29<_psychic_>but it *is* social.
17:40:01<PovAddict>sure
17:52:45<mattl>we need an anti-social network.
17:53:52<mattl>in which every time you befriend someone, it blocks you from seeing all their activity, etc.
17:54:35<Anchakor>mattl: funny, but maybe useful as blacklist sharing :)
18:01:35<Shepard>https://jetpack.mozillalabs.com/
18:01:40<Shepard>H:|Mozilla Labs Jetpack
18:02:00<Shepard>H:"Jetpack is an API for allowing you to write Firefox add-ons using the web technologies you already know."
18:02:29<LeeF>As opposed to what? things like XUL?
18:02:38<PovAddict>I guess
18:03:03<Shepard>H:See also the [blog post about it|http://ajaxian.com/archives/jetpack-extensions]
18:05:09<Anchakor>wasn't mozilla the one developing XUL?
18:05:19<PovAddict>omg
18:05:38<PovAddict>I just found a JS library that helps work around problems with extending Object.prototype and for..in
18:05:47<PovAddict>"The inClean method returns a modified version of your function, where it makes sure that all for-in-loops and if-ins does not detect Object.prototype members."
18:06:27<PovAddict>all nice until i saw the implementation: it calls toString on your function to get the code, applies regex replacements to the code, then compiles it again with eval
18:31:14<tommorris>http://gist.github.com/115594
18:31:38<tommorris>I:|Some experiments in Jena on JRuby
18:33:49<tommorris>I: I like the idea of using JRuby and Jena to build Semantic Web shells. Not quite sure what for, yet, but I'm thinking that JRuby may be the platform I use when I next come back to faffing with Reddy.
18:36:27<tommorris>I: (Now that Nokogiri, the libxml-based XML/HTML parsing library, works with the bleeding-edge version of JRuby using FFI, one could write a nokogiri-based RDF/XML - or RDFa, eRDF, microformats, bla bla bla - and have it work on both JRuby and MRI. Then one can write fun little SemWebby hacks in JRuby, hack at them in the jirb shell and compile 'em into .class files.)
18:37:51<kidehen>G: |Virtuoso Sponger Cartridge XSLT components
18:37:56<tommorris>I: I'm hoping to write a tutorial on writing lightweight scutters soon using JRuby, Nokogiri, Mulgara and possibly Jena/ARQ.
18:39:37<tommorris>G:|Virtuoso Sponger Cartridge XSLT components on GitHub
18:39:44<Anchakor>tommorris: semantic web shells as like BASH, KSH... ?
18:40:54<tommorris>Anchakor: no, more like a place where you could basically load up some RDF and prod at it programatically, without having to, oh, fire up a text editor and write some code
18:42:11<tommorris>Ruby has a strong culture of domain-specific languages - see http://weblog.jamisbuck.org/2006/4/20/writing-domain-specific-languages/
18:43:22<tommorris>The point about them is that they *look* like they are something clever, but in fact they are just Ruby.
18:44:00<tommorris>so a domain-specific shell would just be an interactive Ruby shell with some simple methods for loading RDF and poking at it.
18:45:37<Anchakor>ok
18:52:42<tommorris>kidehen: Just looking at the OpenLink RDFizers - I'm guessing these are done using the HTML representations of the resources
18:53:52<tommorris>What about if one needs to load in, say, two or three XML or HTML resources and transform them all to produce an RDF/XML description of a particular resource?
18:54:00<tommorris>Oh, okay. Seeya kidehen. ;)
18:54:42<PovAddict>heh
18:56:54<dajobe>seems like AtomOwl is pretty close to what I have for my atom-rdf experiments
19:13:44<idmclean>Greetings Semanticians. I hope your pursuits are fruitful.
19:15:21<idmclean>What's new in the semantic web?
19:17:22<Anchakor>:) http://planetrdf.com/
19:28:29<idmclean>Anchakor, what do you need in an ontology to make your magic work with classic programming languages?
19:31:16<Anchakor>idmclean: parser error
19:31:59<idmclean>Anchakor, there is a parser error generated from the ontology side?
19:32:18<Anchakor>idmclean: I meant I failed to parse your question :)
19:39:01<idmclean>Inside the ontology graph you made, you have defined abstract concepts, processing instructions, snippets, and programming language. Could you elaborate on these, please?
19:40:52<Anchakor>idmclean: well the basis is the snippets - Trish is basically a template processor (but a bit more)
19:42:49<idmclean>Anchakor, in this case the template is defined by the ontology, yes?
19:43:08<Anchakor>idmclean: so the snippets are pieces of code with semantic information, which are used to build a program in classic programming languages (like playing with Lego), which can be then executed
19:51:23<tommorris>ACTION really loves Jena, but really loathes Java stack-traces
19:55:25<Anchakor>idmclean: the templates are the processing instructions which are a lists of snippets and empty holes for other processing instructions
19:55:31<mhausenblas>LeeF around?
19:57:07<mhausenblas>hm. what about kasei then? ;)
19:57:09<idmclean>Anchakor, you would need to define templates for the respective programming languages?
19:57:37<mhausenblas>just wanted to query you (LeeF or kasei, whoever is first ;) re http://lists.w3.org/Archives/Public/public-rdf-dawg/2009JanMar/0186.html
19:58:27<mhausenblas>we've restarted work and compiling a list of features for voiD 2.0 and SPARQL-related stuff is quite high up ...
19:58:29<Anchakor>idmclean: yes, template = snippet with holes to fill in
19:58:47<mhausenblas>so, would like to discuss use cases and requirements from your side
19:59:23<mhausenblas>lemme know when you're around, go to catch some sleep now
19:59:35<mhausenblas>phenny, tell LeeF see http://chatlogs.planetrdf.com/swig/2009-05-21.html#T19-55-44
19:59:35<phenny>mhausenblas: I'll pass that on when LeeF is around.
19:59:59<mhausenblas>good nite Web of Data
20:00:11<kasei>hello?
20:00:14<kasei>somebody mentioned my name?
20:00:26<PovAddict>[16:57] <mhausenblas> just wanted to query you (LeeF or kasei, whoever is first ;) re http://lists.w3.org/Archives/Public/public-rdf-dawg/2009JanMar/0186.html
20:00:43<PovAddict>but he's gone now
20:00:48<kasei>alas
20:01:56<kasei>phenny, tell mhausenblas email or twitter is the best way to reliably ping me. very interested to talk about void2.
20:01:56<phenny>kasei: I'll pass that on when mhausenblas is around.
20:03:51<idmclean>Anchakor, do you have any templates developed that I could use as a sample?
20:04:09<idmclean>or formal requirements?
20:06:36<Anchakor>idmclean: no, I am now thinking about data types which are used on input/output of the processing instructions
20:08:06<idmclean>Anchakor, let's apply Occam's razor to the problem. What is the smallest, simplest, least expressive programming language you comprehend?
20:09:45<Anchakor>idmclean: does Galculator's paper mode count? :)
20:10:25<Anchakor>idmclean: http://en.wikipedia.org/wiki/Karel_(programming_language) :)
20:12:38<idmclean>Anchakor, can Karel be executed and evaluated on commandline?
20:13:35<idmclean>Never mind about that part. It seems well defined enough for us to build a simple template.
20:33:54<LeeF>phenny, tell mhausenblas be happy to talk, who is "we" that has restarted work?
20:33:55<phenny>LeeF: I'll pass that on when mhausenblas is around.
20:33:56<phenny>LeeF: 19:59Z <mhausenblas> tell LeeF see http://chatlogs.planetrdf.com/swig/2009-05-21.html#T19-55-44
20:33:58<idmclean>Anchakor, what would we need to develop to make a viable Karel template?
20:33:59<LeeF>thanks, phenny
20:34:52<Anchakor>idmclean: I say forget Karel, let me build an example with the ontology draft I have in mind and lets work with that
20:35:15<idmclean>Anchakor, elaborate, please.
20:38:18<Anchakor>idmclean: well I have some thought about the ontology already, but it's too dificult to put on paper now, so Ill try to make a diagram and an example
21:28:26<Anchakor>just reading about freebase exposing linked data... are their uris persistant?
22:22:43<kidehen>Anchakor: yes they are :-)
22:28:53<Anchakor>oh nice, than that might make freebase the ideal referencing db with all those sameAs links I heard they provide...
22:29:54<Anchakor>if there are no snags, this is really really cool
22:31:53<kwijibo>kidehen: you have followed the voiD stuff - what is your view on this? I am trying to explain the difference between a sparql dataset and a void Dataset. As I see it, a sparql dataset is something defined by the consumer of data, and consists of graphs. A voiD Dataset is something defined by the publisher (others can add statements about the dataset, but not to it)
22:32:48<kwijibo>and is not defined in terms of graphs
23:19:09<kidehen>phenny, tell kwijibo: voiD Data set is typically produced by the Triple / Quad Store instance
23:19:09<phenny>kidehen: I'll pass that on when kwijibo is around.
23:20:13<kidehen>phenny, tell kwijibo: voID data set takes the form of a graph e.g. http://lod.openlinksw.com/void/Dataset
23:20:14<phenny>kidehen: I'll pass that on when kwijibo is around.
23:46:42<idmclean>phenny, tell tobyink check out prototype theory, frame semantics, mental spaces, construction grammars, word grammars, and cognitive grammars. Also, The chomsky hierarchy is a dialect of a language.
23:46:42<phenny>idmclean: I'll pass that on when tobyink is around.

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