Experimental IRC log swig-2009-05-28

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.

02:40:05<adu>hey
02:41:43<adu>What is the relationship between rdf:List types and xsd:anySimpleType types? Are they both subclasses of the same underlying List type?
05:31:22<mhausenblas>good morning Web of Data
06:21:38<mhausenblas>http://www.google.com/webelements/
06:21:51<mhausenblas>A:| Big G's Web Elements
06:22:36<mhausenblas>A: wondering what it would take to drive them by RDF
07:43:05<phrearch>hi
07:44:47<phrearch>when using sparql with php, should i go for arc or?
07:58:40<mhausenblas>phrearch: yes
07:59:26<phrearch>where does owl fit in the picture?
08:00:26<phrearch>with sparql i could query owl as well i guess
08:06:47<mhausenblas>phrearch: you always query RDF, right?
08:07:31<mhausenblas>OWL is one way (maybe not the best way to start with) to define the terms of your domain at hand
08:07:31<phrearch>got a lot of stuff to read up about, before i slightly understand this semantic stuff :)
08:07:51<mhausenblas>got XML or RDBMS background?
08:08:17<phrearch>yes, im working on an opensource webinterface
08:08:38<phrearch>but there are alot of good practices i still need to learn about
08:09:03<mhausenblas>ok, so you know the diff between, say an XML Schema and an XML document right? or between an RDB schema and the actual cells in a table, yes?
08:09:53<phrearch>ehm, the schema describes how the document should be structured right?
08:10:02<mhausenblas>yes, same here. you have schema languages such as RDF Schema or OWL, FWIW
08:10:10<mhausenblas>and the instance level
08:10:26<mhausenblas>in RDF you actually don't think in terms of documents in the first place
08:11:03<mhausenblas>you think resources, entities, nodes or whatever, it's a graph structure, that's the point ;)
08:11:16<phrearch>so, with sparql, a query should find its own way through a web of rdf/owl i guess?
08:11:40<phrearch>yea, i really like that idea
08:11:46<mhausenblas>SPARQL is the same to RDF what SQL is to relational database table(s)
08:11:58<phrearch>im trying to imagine how new generation public virtual worlds fit into this picture
08:12:13<phrearch>i see
08:12:20<mhausenblas>very little to do with the schema, actually. simple pattern matching
08:12:37<mhausenblas>got some more 5min so we can walk through a simple example if you want ...
08:12:49<phrearch>that would be great
08:13:21<mhausenblas>ok, so what is your favourite sport, book or whatever
08:13:29<phrearch>ehm snowcrash
08:13:41<phrearch>its a book about virtual worlds
08:13:52<mhausenblas>ok, let's see if it's in dbpedia ;)
08:14:22<phrearch>dbpedia is wikipedia with rdf support ?
08:14:34<mhausenblas>sort of, yes ;)
08:14:37<mhausenblas>ACTION looking at http://lookup.dbpedia.org/
08:14:52<mhausenblas>1. step: get into the RDF world
08:15:06<phrearch>heh, it returns the data
08:15:13<mhausenblas>so I'm using a lookup service that takes a term ('snowcrash')
08:15:23<mhausenblas>and gives me resources that talk about this
08:15:25<mhausenblas>let's see
08:15:40<mhausenblas>so, is it http://dbpedia.org/resource/Snow_Crash ?
08:16:04<phrearch>yes :)
08:16:15<mhausenblas>ok, next step. we want to learn more about it, right?
08:16:29<phrearch>yea, like genre and stuff
08:16:39<mhausenblas>now, I go to the public SPARQL endpoint of DBpedia
08:16:44<mhausenblas>which is at http://dbpedia.org/sparql
08:17:17<phrearch>ok
08:17:29<mhausenblas>I enter following query:
08:17:36<mhausenblas>select distinct * where { <http://dbpedia.org/resource/Snow_Crash> ?p ?o }
08:18:19<mhausenblas>which basically says: the subject (first position, is fixed with the URI that identifies the resource that talks about snow crash, right?)
08:18:43<mhausenblas>and ?xxx denotes that xxx is a variable
08:18:58<mhausenblas>so what do you think I'm asking for :)
08:19:25<phrearch>ehm, search for all Snow_Crash related stuff and store it in ?p and ?o ?
08:19:33<mhausenblas>yep, sort of
08:19:50<mhausenblas>we' say bind it to ?p and ?o but other than this, perfect :)
08:20:10<mhausenblas>ok, so what are we interested in (looking at the ?p column?)
08:20:10<phrearch>distinct is more strict?
08:20:18<mhausenblas>removes duplicates
08:20:22<phrearch>ah ok
08:20:39<phrearch>ehm, whats in ?p and whats in ?o
08:20:44<phrearch>arent they simular?
08:20:50<mhausenblas>?p is the property
08:21:10<mhausenblas>relating two resources OR relating a resource to a literal value
08:21:12<phrearch>aha, they are predefined vars to use
08:21:15<mhausenblas>no
08:21:21<mhausenblas>you can use whatever you want
08:21:44<mhausenblas>select distinct * where {
08:21:44<mhausenblas><http://dbpedia.org/resource/Snow_Crash> ?IAMARELATION ?IMAVALUE
08:21:44<mhausenblas>}
08:21:46<mhausenblas>same
08:22:14<mhausenblas>just makes sense to give it nice names (same as with selfspeaking vars in a programming language, isn't it? :)
08:22:22<phrearch>yea :)
08:22:46<mhausenblas>but for now I just wanted to dump *all* props and *all* objects
08:22:51<mhausenblas>so it didn't matter
08:22:57<mhausenblas>now, what are we after?
08:23:29<phrearch>ehm, i would like to know all resources in dbpedia mentioning the word snowcrash. is that possible?
08:23:35<mhausenblas>side note: literals (such as a string "XXX" or a number 123^^xsd:int) are finie
08:23:57<mhausenblas>phrearch: that's basically the lookup service we used in the first step
08:24:04<mhausenblas>but know we are in the RDF world
08:24:23<mhausenblas>we are exploring the 'semantic space' by following links
08:24:41<mhausenblas>these links are the properties between resources (the ?p, above)
08:25:11<mhausenblas>so, as I said, literals a fine (for humans to look at, displaying, driving a sensor or whatever)
08:25:50<mhausenblas>but the IMO interesting part is when the value (?o) is again a resource that can be a subject in another statement
08:26:02<mhausenblas>ok, too much I guess. let's step back, ok? :)
08:26:13<phrearch>ehm, trying to comprehend it
08:26:16<mhausenblas>two concrete examples,right?
08:26:31<phrearch>like you do one query, then query ?p again for new results of another resource?
08:26:39<mhausenblas>yep, sort of
08:26:46<mhausenblas>so, for example:
08:26:53<mhausenblas>what is the title of the resource?
08:27:19<phrearch>Snow Crash
08:27:22<mhausenblas>translates for example to http://www.w3.org/2000/01/rdf-schema#label
08:27:29<mhausenblas>yes, I know that you know it :)
08:27:42<mhausenblas>but what would the SPARQL query look like? :)
08:28:31<mhausenblas>select distinct * where { <http://dbpedia.org/resource/Snow_Crash> <http://www.w3.org/2000/01/rdf-schema#label> ?o }
08:29:37<phrearch>ah, by referencing to that rdf-schema, you would know which properties an rdfs can have?
08:29:46<mhausenblas>which you can perfectly read as "I want to know of the dbpedia:Snow_Crash resource the rdfs:label ..."
08:29:52<phrearch>or is this just one example of a rdfs
08:30:09<mhausenblas>yeah, sort of
08:30:22<mhausenblas>you need to be aware of the vocabulary (schema), yes
08:30:50<phrearch>ow nice
08:30:52<mhausenblas>so I could have tried dcterms:label
08:31:00<phrearch>it returns the title in like 8 languages
08:31:19<mhausenblas>if you're unsure about the prefixes (rdfs:, dcterms: etc.) go look it up at prefix.cc
08:31:41<mhausenblas>this gives you the most common usages (like more than 90% they are used in this way)
08:31:51<phrearch>interesting technique
08:32:00<mhausenblas>so rdfs: usually goes with <http://www.w3.org/2000/01/rdf-schema#> right?
08:32:03<mhausenblas>yeah :)
08:32:11<mhausenblas>but this was the uninteresting part, IMO
08:32:14<mhausenblas>just a literal
08:32:21<mhausenblas>you can view it, that's it
08:32:40<mhausenblas>(ok, there are cases you can do more with it, say if it's a number or a date or whatever)
08:32:51<mhausenblas>but back to the interesting part, right?
08:32:54<phrearch>yea :)
08:33:06<phrearch>this has been pretty amazing so far though
08:33:20<mhausenblas>hehe. ok, so we are interested in an object (?o) that is not a literal
08:34:39<mhausenblas>right so I say in the query that I don't want ?o to be a literal
08:34:48<mhausenblas>select distinct * where { <http://dbpedia.org/resource/Snow_Crash> ?p ?o FILTER isURI(?o) }
08:35:02<phrearch>hm, lets try
08:35:16<mhausenblas>now, this is not entirely true (cause there are also so called bNode, blank nodes)
08:35:43<mhausenblas>which are also resources that can play a role that is be a subject of a statement
08:35:48<mhausenblas>but they are not so nice
08:35:58<mhausenblas>they are sort of local to an RDF graph
08:36:20<mhausenblas>which means you can't refer to it from outside and hence they provide no 'entry point' to your data
08:36:35<reto>that's why ifp's are for
08:36:43<mhausenblas>best is to avoid them (in the creation) but this is not always possible
08:36:50<mhausenblas>ta reto ;)
08:37:00<reto>disagree with avoiding
08:37:03<phrearch>so, that last query returned all resources involved with the title?
08:37:30<mhausenblas>reto: yes, but let's not confuse phrearch right no, I'm trying to intro him to SPARQL/RDF, ok?
08:37:31<reto>you such give uris only to thing for ehich you vcan assume that there aren thound others giving a different uri to the same thing
08:37:58<reto>ok
08:38:10<phrearch>ACTION is a noob
08:38:12<mhausenblas>no, phrearch, same as the intial query, just restricting the objects to the 'interesting' ones
08:38:20<mhausenblas>reto: thanks :)
08:38:30<mhausenblas>phrearch: that's why we do it step by step :)
08:38:39<mhausenblas>ok,so we can pick one
08:38:46<phrearch>so, suppose i want to know where the author is coming from?
08:39:01<mhausenblas>good question!
08:39:02<reto>ACTION wonders what the statement "bnodes are not so nice" has to do in a sparql intro
08:39:17<phrearch>neil stephenson is an american i think
08:39:24<reto>ACTION passing the border to austria
08:39:33<mhausenblas>hehe
08:39:42<mhausenblas>ok, so phrearch slow down
08:39:51<phrearch>ok :)
08:40:01<mhausenblas>what relates dbpedia:Snow_Crash to its author?
08:40:31<phrearch>ehm, its in the rdfs right?
08:40:42<mhausenblas>well, sort of ;) maybe the <http://dbpedia.org/property/author> property?
08:40:51<mhausenblas>ACTION looking at the ?p column
08:40:59<mhausenblas>let's try it
08:41:17<mhausenblas><http://dbpedia.org/resource/Snow_Crash> <http://dbpedia.org/property/author> ?o
08:41:42<mhausenblas>or, as we now expect an author as the object, I could have also written ?author
08:41:48<phrearch>http://dbpedia.org/property/author
08:42:12<mhausenblas>ok, did you try the query above?
08:42:19<mhausenblas>what's your result?
08:42:52<phrearch>ill be damned
08:42:56<mhausenblas>hu?
08:43:03<mhausenblas>why? :)
08:43:06<phrearch>ah no. its neil stephenson
08:43:08<reto>ACTION disconnects to avoids roaming charges, and hopes to see some of you at the http://www.iks-project.eu meeting
08:43:14<phrearch>:)
08:43:23<mhausenblas>ACTION waves bye to reto - success!
08:43:29<mhausenblas>so my answer is http://dbpedia.org/resource/Neal_Stephenson
08:43:42<reto>ACTION waves back
08:43:45<phrearch>yea, which is $author
08:43:48<phrearch>ehm ?author
08:44:11<mhausenblas>yeah, $author is also ok (sort of legacy ;)
08:44:20<mhausenblas>but do use ?xxx
08:44:42<mhausenblas>ok, now we have a new resource, namely the one identified by <http://dbpedia.org/resource/Neal_Stephenson> ok?
08:44:47<mhausenblas>so, we can do the same again
08:45:07<phrearch>interesting, so you then query this new resource, for things like country
08:45:10<mhausenblas>learn about the resource
08:45:11<mhausenblas>yes
08:45:14<mhausenblas>e.g. <http://dbpedia.org/resource/Neal_Stephenson> ?p ?o
08:45:53<mhausenblas>so, you wanted to know where he was born? or when?
08:46:09<phrearch>ehm, where
08:46:27<mhausenblas>ok, executing the above query which ?p would you oick?
08:46:35<mhausenblas>s/oick/pick
08:47:10<phrearch>http://dbpedia.org/property/birthPlace
08:47:16<mhausenblas>hehe, yes
08:47:24<mhausenblas>I guess you've got it ... :)
08:47:48<mhausenblas>and now, final one: how would the query look like?
08:48:24<mhausenblas>precisely, yes, it's <http://dbpedia.org/resource/Neal_Stephenson> <http://dbpedia.org/property/birthPlace> ?o
08:48:25<phrearch>ehm, select distinct * where { <http://dbpedia.org/resource/?author> ?p ?o }
08:48:31<mhausenblas>hehe
08:48:43<mhausenblas>we were at the author already, right?
08:48:58<phrearch>ah, you travel along with each resource behind the other
08:49:03<mhausenblas>yep
08:49:46<phrearch>thats really amazing stuff :)
08:49:56<mhausenblas>basically follow-your-nose with SPARQL ;)
08:50:28<mhausenblas>http://inkdroid.org/journal/2008/01/04/following-your-nose-to-the-web-of-data/
08:51:00<mhausenblas>D: Follow-your-nose (FYN)
08:51:40<mhausenblas>phrearch, btw, you can have a log/bookmark of our conversation at http://swig.xmlhack.com/
08:52:14<mhausenblas>ok, gotta head out now
08:52:28<phrearch>ok, thanks for explaining this
08:52:58<mhausenblas>if you have any more questions just ask here - there are a lot (also more competent) people hanging out here
08:52:59<phrearch>this will be fun to experiment with :D
08:53:11<mhausenblas>yep, enjoy!
08:53:28<phrearch>thanks again. later ;-)
08:53:33<mhausenblas>(btw, if you want to leave me a message just use 'phenny, tell mhausenblas XXX')
08:53:38<mhausenblas>ok, cya
08:53:55<phrearch>ok
09:01:09<phrearch>hm, suppose i would like to search for the birthplace of the author of the book snowcrash
09:01:23<phrearch>i know i can find the birthplace when knowing the author
09:01:24<phrearch>select distinct * where {<http://dbpedia.org/resource/Neal_Stephenson> <http://dbpedia.org/property/birthPlace> ?o}
09:12:23<Anchakor>Phae: select distinct ?birthPlace where {<http://dbpedia.org/resource/Snow_Crash> <http://dbpedia.org/property/author> ?author . ?author <http://dbpedia.org/property/birthPlace> ?birthPlace . }
09:12:30<Anchakor>phrearch: ^
09:12:55<phrearch>cool. gonna try that
09:12:57<Anchakor>should be it (I hope so)
09:13:41<phrearch>yea, that worked fine
09:14:03<phrearch>$author and $birthplace are the properties?
09:14:21<phrearch>and the dot is to pass on the info or?
09:14:44<Anchakor>no, property is always in the middle of triple (unless you are defining it with rdfs/owl)
09:15:19<Anchakor>the dot means that next thing coming is another full triple pattern for the parser
09:15:22<Phae>:O
09:17:21<Anchakor>Phae: sorry, my bad :) that was for phrearch obviously
09:17:37<phrearch>would it be possible to query geonames where the author's birthplace is?
09:17:55<phrearch>like in lat/lon
09:20:51<Anchakor>dunno if this information is in dbpedia... it is still dificult querying multiple datasets now
09:26:06<phrearch>are there any cross-webservice examples?
09:30:52<Anchakor>no, the technology is called federated query, but it's still under development, it's pretty advanced stuff
09:31:38<Anchakor>anyway I gtg... bes what can you read is http://www.w3.org/TR/rdf-sparql-query/ of course, but this can be good for basics http://jena.sourceforge.net/ARQ/Tutorial/basic_patterns.html
09:36:22<phrearch>ok, thanks
10:13:59<phrearch>hm, federated queries use DARQ?
10:14:20<phrearch>ah arq
10:14:44<shellac_>DARQ is a particular federated query system that uses ARQ
10:15:35<phrearch>so, the basic technology behind queries between fi geonames and dbpedia is called arq?
10:16:22<shellac_>no, ARQ is a piece of software that implements SPARQL (and extensions thereof)
10:16:46<shellac_>I think dbpedia uses virtuoso
10:17:11<gromgull>Can someone give me a good rdfa example page? Which is NOT ivan herman's? :)
10:17:25<karlcow>http://twitterdata.org/
10:17:48<IvanHerman>:-(
10:17:50<karlcow>E:|Twitterdata - definition of a syntax for twitter data
10:17:59<shellac_>that's a bit harsh :-)
10:18:06<libby>rdfa example page of what gromgull?
10:18:17<gromgull>of anything ... a page with rdfa embedded
10:18:26<gromgull>Sorry Ivan - I've already tested your page :)
10:18:37<libby>njh has a great homepage
10:18:43<IvanHerman>:-)
10:18:51<libby> http://www.aelius.com/njh/
10:18:58<libby>(with rdfa I mean)
10:18:58<shellac_>slide share, digg have simple stuff
10:19:09<karlcow>I wonder if rdfa.info collects pages of commercial oriented sites containing rdfa
10:20:15<shellac_>ACTION tries to remember which pages Jeni did for London Gazette
10:20:31<gromgull>excellent - thanks liiby and shellac!
10:20:34<gromgull>libby
10:23:20<shellac_>gromgull: example for you http://www.london-gazette.co.uk/issues/59069/notices/824811/recent=10;category=transport;subcategory=highways
10:24:55<gromgull>Thanks again - nice example
10:25:09<gromgull>For the interested I am testing the sparks ozone browser: http://www.dcs.shef.ac.uk/~gregoire/sparks/
10:25:35<gromgull>it's doesn't have many plugins for the kind of things on that page though ... in fact, it really works best on Ivan's page :)
10:28:04<IvanHerman>gromgull: I have a more complex one for you, but it is still mine: http://www.ivan-herman.net/foaf.html
10:28:29<IvanHerman>a simple, but nice example (and this is not mine:-) http://id.loc.gov/authorities/sh91000664
10:30:59<gromgull>nice - again the sparks browser only really supports events and twitter profiles though...
10:31:04<gromgull>How is the Semanitc Radar these days?
10:31:10<gromgull>it doesn't work with my firefox beta
10:32:31<melvster>gromgull: http://sw-app.org/mic.xhtml#i and http://tobyinkster.co.uk/ also there's the libre.fm stuff
10:40:46<mhausenblas>gromgull: go to any slideshare or digg page :)
10:40:56<mhausenblas>ACTION off again :)
10:41:55<melvster>or: http://alpha.libre.fm/artist/U2
10:44:19<gromgull>many of these fail with "Sorry, jqueryrdfa does not support inline definition of CURIE at the moment."
10:44:30<gromgull>not the fault of the pages of course
11:19:52<kwijibo>fgiasson: as mhausenblas said yesterday, when cygri, mhausenblas , jun and I were writing the guide and vocab, we did consider minting a new property instead of using dct:isPartOf but we couldn't think of a justified reason for it, so we went with dct:isPartOf
11:20:30<kwijibo>so if we are to reconsider, we really need some reasons to evaluate
11:21:34<kwijibo>I think yesterday you said that it is harder to query if generic predicates are used
11:22:16<kwijibo>but doesn't specifying the types of the nodes in the query answer this objection ?
12:20:35<melvster>excellent #swig works in pidgin ... along with pretty much everything else
12:34:41<gromgull>Anyone here already in Crete?
12:34:46<gromgull>For STI offsite f.x. ?
12:39:28<kwijibo>mhausenblas: cygri I think the dct:isPartOf thing relates to something danbri said recently in a mailing list thread about "these things will always be a bit fuzzy until our vocabularies have something like the ideas behind FRBR built into them"
12:56:23<cygri>kwijibo: yeah i have to work through these 200 unread public-lod mails :-(
12:58:49<cygri>FRBR for talking about web data? interesting. wikipedia as a creative endeavour is a WORK, dbpedia an EXPRESSION of it, the dbpedia sparql endpoint a MANIFESTATION, and a particular response an ITEM???
13:18:19<gromgull>Any talis people around?
13:19:15<BenO>gromgull, have you tried in #talis
13:19:17<BenO>?
13:19:23<gromgull>Good point :)
13:19:27<gromgull>\join #talis
13:19:30<gromgull>damn irc skillz
13:32:10<ldodds>gromgull: hi, whats the q?
13:32:24<gromgull>ldodds - i got help in #talis! Thanks
13:32:48<ldodds>heh, picked the wrong channel to catch up on first
13:32:48<gromgull>I was after a hi-res logo - rjw is asking design people for me
13:32:51<ldodds>ok
13:33:31<kwijibo>interesting example cygri
13:33:31<BenO>ldodds, heh, I can see where your loyalties lie ;)
13:34:35<ldodds>:)
13:34:36<cygri>kwijibo, i made it up without thinking, but this might be the general idea
13:35:05<ldodds>ACTION not sure about cygri's FRBR classification of wikipedia/dbpedia
13:35:22<cygri>ACTION not sure either
13:35:25<ldodds>:)
13:36:43<kwijibo>maybe, as applied to documents-as-graph, the Expression is the graph, and the Manifestation is the document serialising it? The item would be the copy you download when deref the URI
13:36:59<cygri>not sure if the WORK/EXPRESSION/MANIFESTATION etc distinction is the right one in this context
13:37:22<cygri>but yeah that sounds about right kwijibo
13:38:17<kwijibo>yeah, in general because there are no URIs for either the item or the Expression
13:40:09<ldodds>I suspect danbri was referring to the fact that most RDF modelling doesn't take FRBR constructs into account so, for example, when we refer to a "Document" what are we referring to? Work, Manifestation, etc
13:40:18<kwijibo>yeah
13:40:24<kwijibo>i think it was along those line
13:40:27<kwijibo>*lines
13:40:44<ldodds>which is important, as we often mix concerns currently.
13:41:02<ldodds>been meaning to blog something about that w.r.t wishlists, etc, illustrate the issues
13:41:25<kwijibo>well, i agree that we mix concerns currently - but is it important ?
13:41:34<ldodds>yes, it can be very important
13:41:59<kwijibo>what would an example be showing the importance?
13:42:43<ldodds>for example we could have a conversation about "Lord of the Rings", we don't need to worry about FRBR issues there as for the purposes of our conversation we're more than likely discussing the Work
13:43:20<ldodds>but if I add something to a wishlist, and you're shopping for a present for me, then do I just want "Lord of the Rings" or do I want a specific manifestation of that book?
13:44:05<ldodds>at the moment we don't typically distinguish between this kind of thing
13:44:35<ldodds>for certain types of scholarly research, its important to distinguish between the item levels, e.g. to compare actual instances of a manuscript
13:45:34<ldodds>if I'm reviewing a film, am I reviewing a specific cut (manifestation) or the work?
13:45:48<ldodds>expression, sorry
14:00:15<kwijibo>ldodds: yeah, sure, I'd agree with that - we were talking about that yesterday with the man from cambridge - sometimes people want to specify an edition, sometimes they don't care
14:00:44<kwijibo>(when building reading lists)
14:00:54<ldodds>yep
14:01:29<kwijibo>is there a common fudge of these concepts thought that has undesired consequences though?
14:01:32<kwijibo>*though
14:01:40<ldodds>its my main gripe with a lot of book social networking sites -- they generally use amazon apis, and I end up reviewing, say, the paperback edition of "Matter" which exist separately to hardback
14:01:58<kwijibo>yeah, that's true
14:02:40<kwijibo>and you think you've told it you've read such and such a book, and then it asks you if you've read a different edition
14:02:48<ldodds>Not sure what a fudge would be -- there's problems whether you assume too coarse or too fine-grain an approach
14:02:50<ldodds>ACTION nods
14:02:52<kwijibo>and you realise you've said something more specific than you intended
14:03:10<kwijibo>and you end up looking like you've read the same book 5 times in different editions
14:03:43<kwijibo>but that's an application problem isn't it ?
14:06:02<ldodds>its a problem for anyone using someone else's data
14:14:06<kwijibo>ldodds: well, to be more specific, do you think there is a practical problem with treating a document as a graph ?
14:14:46<ldodds>sounds like distinct things to me
14:16:20<ldodds>kwijibo: what was the use case, I obviously missed earlier discussion
14:16:46<kwijibo>ldodds: I agree that they are, but i'm undecided about whether it matters
14:17:35<ldodds>is this in context of describing contents of a Void dataset?
14:18:03<ldodds>have you looked at OAI-ORE? That has some terms and modelling around describing collections of things. Might be useful
14:18:48<kwijibo>ldodds: yes, it's inthe context of voiD, but its; sort of already based on an established pattern - using the document uri to talk about the collection of triples represented by the document
14:19:03<kwijibo>or maybe i should say "serialised by the document"
14:24:35<kwijibo>ldodds: cygri - i think the document-as-graph idiom comes from SPARQL ? eg, if you use http://sparql.org/sparql.html you can specify document uris in the FROM clauses, and they are treated as graphs, aren't they ?
14:26:17<ldodds>FROM defines a graph, doesn't say anything about documents or how the sparql engine maps the uri to data
14:28:21<kwijibo>ACTION rereads SPARQL spec
14:28:52<kwijibo>ah, it touches on precisely what I'm talking about: The FROM NAMED syntax suggests that the IRI identifies the corresponding graph, but the relationship between an IRI and a graph in an RDF dataset is indirect. The IRI identifies a resource, and the resource is represented by a graph (or, more precisely: by a document that serializes a graph). For further details see [WEBARCH].
14:28:52<shellac_>kwijibo: I suspect it came from the named graphs paper
14:29:38<kwijibo>shellac_: which one ?
14:29:58<shellac_>carroll et al
14:30:39<shellac_> http://www.hpl.hp.com/techreports/2004/HPL-2004-57R1.html
14:31:27<kwijibo>ah thanks
14:33:29<CaptSolo>hi all
14:36:10<cygri>kwijibo, ldodds: glossing over the difference between "named RDF graph" and "RDF document with a URL" has a long tradition, we've been doing this in NG4J for many years (long before SPARQL)
14:36:15<cygri>and it works for me
14:36:22<kwijibo>;)
14:36:55<cygri>the only issue i've seen is with versioning... a web document can serialize different graphs over time
14:36:56<kwijibo>in REST terminology, is the named graph the resource and the document the representation ?
14:37:07<cygri>kwijibo, no
14:37:17<cygri>the document is the resource
14:37:47<cygri>the named graph doesn't show up in REST terms
14:37:49<kwijibo>so, there are more things than there are uris for things
14:37:55<cygri>because REST doesn't have a data model
14:40:11<ldodds>it does to the extent that it defines Resources and Representations
14:40:17<kwijibo>versioning isn't specific to RDF though ?
14:40:36<kwijibo>atom feeds and html pages and all kinds of documents change over time
16:59:23<MacTed>kwijibo - There will always be more URIs-of-things than there are things-which-have-URIs, because it is impossible for everyone who wants to Name a Thing to discover all pre-existing Names for that Thing.
17:00:17<kwijibo>MacTed: I was saying the opposite (within specific context, not in the whole universe)
17:00:55<MacTed>kwijibo - yes, I saw you say the opposite. perhaps I'm not understanding your specific context, but I am pretty sure that I disagree. :-)
17:01:12<kwijibo>because RDF documents represent graphs, and the graphs don't have names
17:01:24<kwijibo>so the document has a name, but the graph doesn't
17:01:29<MacTed>cygri - both a Document and a NamedGraph are Resources. the Document which is a Representation of a NamedGraph is *also* itself a Resource.
17:01:46<MacTed>why don't the graphs have names?
17:02:09<cygri>MacTed: "rdf graph" as defined in the RDF spec doesn't have a name, as such
17:02:22<cygri>of course, you can name anything, so you can give it a name
17:02:29<MacTed>my point exactly
17:02:33<kwijibo>dunno, you can give them names i suppose, but there isn't really an arhcitecture behind it
17:03:26<cygri>MacTed: however, a "document" in the sense of "web page" is *not* a representation
17:03:36<MacTed>cygri - I disagree.
17:03:41<cygri>(using representation in the REST sense)
17:04:09<PovAddict>I think it's more like: this bunch of HTML is a representation of the document
17:04:10<MacTed>you cannot GET a Resource. you can only GET a Representation of that Resource.
17:04:35<cygri>PovAddict: correct
17:04:43<kwijibo>ACTION nods so far
17:04:44<cygri>MacTed: you can perform a GET operation on a resource
17:04:55<PovAddict>the document itself (a bit of an abstract concept) is not a representation
17:04:57<cygri>the result of that operation is a representation of the resource
17:05:05<cygri>PovAddict: exactly
17:05:29<cygri>a lot of confusion comes from the term docuemtn, people use it differently
17:06:21<cygri>sense 1: a web document is a string in a certain format, such as HTML. technical term for this is Representation
17:07:01<cygri>sense 2: a web document is a web page with a URL that i can open in my browser. technical term for this is: information resource
17:07:55<cygri>MacTed: when you talk about Documents vs NamedGraphs earlier, which sense were you using?
17:09:18<kwijibo>ACTION tries to re-understand the sparql spec, again: The FROM NAMED syntax suggests that the IRI identifies the corresponding graph, but the relationship between an IRI and a graph in an RDF dataset is indirect. The IRI identifies a resource, and the resource is represented [...] by a document that serializes a graph
17:10:20<mhausenblas>OpenLink seems to be very popular already ;)
17:10:22<kwijibo>but the usage in the examples suggest that the resource identified by the IRI is /not/ the graph, but the document
17:10:25<mhausenblas>http://linkeddata.uriburner.com/
17:10:39<kwijibo>ouch
17:10:43<mhausenblas>MacTed: do you think you can fix that?
17:11:03<kwijibo>This Site Hacked By :
17:11:03<kwijibo>Persian Boys Hacking Team
17:11:03<MacTed>hrm. lemme see.
17:11:04<kwijibo>Persian Boys Hacking Team From A Land With A History-Long Background
17:11:17<cygri>mhausenblas: was it aftab?
17:13:46<PovAddict>F:0wned site
17:15:06<MacTed>that's not happy. I think I come back to Resources vs Representations vs Documents later.
17:16:29<cygri>+1 MacTed! ttyl
17:16:50<kwijibo>PBHT--
17:18:44<kwijibo>so .. The IRI identifies a(n information) resource, and the IR is represented by a document (sense 1) that serializes a graph ?
17:22:45<cygri>kwijibo, that sounds right to me. although i don't really like the sense 1 of document... i'd say "it's represented by an RDF/XML representation" or "turtle representation" that serializes a graph
17:40:49<kwijibo>does anyone know of a vocabulary that defines a class for an RDF Graph ?
18:19:48<MacTed>F1:""
18:20:20<MacTed>the fun of live-tweaking a test service. someone accidentally left a file world-writeable. oops!
18:20:36<MacTed>not much of a triumph for the Persian Boys Hacking Team
18:23:20<mhausenblas>hehe
18:23:24<mhausenblas>MacTed++
18:25:50<MacTed>*shrugs* I'm more the messenger than the fixer in this case. just alerted those who'd been working on the service recently.
18:28:43<kidehen>MacTed: I have few secs before returning to baby sitting etc..
18:29:05<kidehen>MacTed: who want's to know what? While I can type.
18:31:54<kidehen>cygri: http://linkeddata.uriburner.com/about/html/http/news.cnet.com/2547-1_3-0-20.xml%235
18:33:38<kidehen>mhausenblas: recovery was just about fixing the silly default perms used for the directory that provides root for uriburner domain. It actually had +w for public :-( So deserved to be hacked :-)
18:34:40<kidehen>mhausenblas: so zapped out the piddly html docs by our external QA team (the hackers) and fixed the perms. Then told internal ODS to upstream the home page :-)
18:35:55<kidehen>mhausenblas: recovery time was slightly prolonged on the assumption that a serious infrastructure hack had occurred, but even then we would recover the data space rapidly
18:40:29<Shepard>phenny, tell kwijibo: re. vocabulary for RDF graphs, see http://www.w3.org/2004/03/trix/rdfg-1/ - and also http://www.w3.org/2004/03/trix/
18:40:29<phenny>Shepard: I'll pass that on when kwijibo is around.
18:40:54<mhausenblas>kidehen: thanks for the explanation
18:41:35<kidehen>mhausenblas: if you leave a WebDAV folder with +w and then use that folder to publish you home page, it should be hacked :-)
18:41:39<cygri>good job on the timely repair kidehen and MacTed!
18:42:45<kidehen>cygri: repair was 1) delete the crap (extra $%& for the .asp files) 2) tell ODS to upstream the home page 3) remove +w for public.
18:43:28<kidehen>cygri: if they want to hack us seriously the need to invest a lot more than html and HTTP :-)
18:43:34<cygri>kidehen: no buffer overflow vulnerability in virtuoso? ;-)
18:44:12<kidehen>cygri: shouldn't be, at least not went last passed through our automated QA, but no harm in double double checking.
18:44:46<kidehen>cygri: buffer overrun tests are supposed to be part of all nightly builds, period.
18:48:14<PovAddict>there are clever ways to overflow buffers that not any automated system can test :)
19:34:24<kwijibo>isn't there a REST vocab somewhere?
19:34:24<phenny>kwijibo: 18:40Z <Shepard> tell kwijibo re. vocabulary for RDF graphs, see http://www.w3.org/2004/03/trix/rdfg-1/ - and also http://www.w3.org/2004/03/trix/
19:34:29<kwijibo>does anyone know the uri ?
19:34:34<kwijibo>ta shellac
19:34:40<kwijibo>oops
19:34:42<kwijibo>Shepard
19:34:54<kwijibo>phenny tell Shepard thanks
19:35:05<kwijibo>phenny: tell Shepard thanks
19:35:05<phenny>kwijibo: I'll pass that on when Shepard is around.
19:46:24<mhausenblas>kwijibo: not really (re REST vocab ;)
19:46:30<mhausenblas>we're sort of working on it
19:46:51<mhausenblas>(see http://esw.w3.org/topic/AwwswHome)
19:47:05<mhausenblas>there is however HTTP-in-RDF
19:48:06<mhausenblas>http://www.w3.org/TR/HTTP-in-RDF/
19:48:25<mhausenblas>G:| HTTP in RDF vocabulary
19:49:08<mhausenblas>then there is the Generic Resources Ontology
19:49:10<mhausenblas>http://www.w3.org/2006/gen/ont#
19:49:19<mhausenblas>H: Generic Resources Ontology
19:49:35<PovAddict>weird @ G
19:49:35<mhausenblas>H:| Generic Resources Ontology
19:49:53<mhausenblas>H1: see also http://esw.w3.org/topic/AwwswGenOntDiagrams
19:51:34<mhausenblas>but there is certainly a need for it, so we should deliver it. hopefully soon ;)
19:55:31<kwijibo>http://blogs.talis.com/n2/archives/485
19:55:51<kwijibo>I:| voiD, datasets, graphs, documents, and dcterms:isPartOf backlinks
20:02:06<kidehen>PovAddict: I would suggest you throw some of these at our server, I don't have a problem if you bring it down, its all good QA to me :-)
20:02:42<PovAddict>kidehen: uhh I don't have the skills though :)
20:03:19<PovAddict>in fact I once found a buffer overflow bug in some open source server software and I wanted to try h4xing it before reporting it
20:03:45<PovAddict>I found a tutorial on stack smashing and my head asploded
20:06:51<mhausenblas>ok, off to aharth's farewell party - he's graduating soon and this deserves a Guinness (or two ;)
20:08:02<mhausenblas>(for those who don't know him, he's the guy behind SWSE, Yars, VisNav, etc.)
20:08:07<mhausenblas>good nite Web of Data
20:09:43<kwijibo>night mhausenblas
20:27:34<Anchakor>http://tech.slashdot.org/comments.pl?sid=1248169&cid=28125283
20:27:57<Anchakor>J: interesting in relation to Semantic Web
20:34:58<akozak>I thought one of the big ideas behind the semantic web was machine readability and usefulness, not necessarily creating a "hidden layer" of content.
20:35:27<akozak>re: Anchakor's link
21:50:18<danja>ping
21:50:48<danja>thanks phenny :)
21:57:18<Shepard>hi danja
21:57:18<phenny>Shepard: 19:35Z <kwijibo> tell Shepard thanks
22:05:33<danja>hi Shepard, how goes?
22:06:54<Shepard>not bad :)
22:08:44<danja>I just got a new laptop and apart from aesthetics it's really boring - a computer is just a computer
22:09:52<danja>nicer screen, but that's like getting new specs
22:10:43<danja>a lot faster, but I doubt I'll really notice
22:12:27<danja>gentle kb which is nice (it's a vaio) but then it's a .it kb so it'll take a while to discover the curly brackets
22:13:02<danja>got this symbol though §§§§§§§§§§§
22:14:10<Shepard>heh, so do I §§§ ;)
22:14:32<Shepard>it's quite annoying that programming languages were designed for US keyboards so doing all the brackets is hard finger work ;)
22:14:54<Shepard>for me I mean (.de kb)
22:15:00<PovAddict>well
22:15:09<PovAddict>I have been coding for years on latin american keyboards
22:15:18<PovAddict>I'm so used to it I can't use US keyboards :)
22:15:41<PovAddict>even though my fingers probably suffer the fact that slash is shift-7 when typing file paths
22:16:01<PovAddict>(either that or numpad)
22:16:33<Shepard>well numpad is no good of course if you're using a laptop
22:21:25<danja>fortunately the y is in the right place...
22:22:02<danja>qwertz kbs are a pain
22:22:08<PovAddict>lol
22:22:26<PovAddict>what about trying to use vim in a dvorak keyboard?
22:22:58<Shepard>all a matter of perspective :)
22:23:51<danja>ACTION would rather stick a finger up the dog's bum
22:24:20<Shepard>err, tmi :P
22:25:13<danja>c'mon, either emacs or gedit-type-thing, en-uk kb
22:26:34<danja>ACTION now more used to MS-style text editors
22:51:43<danja>http://logs.jabber.org/jdev@conference.jabber.org/2009-05-28.html
22:52:21<danja>K:| discussions on the Google Wave Jabbery thing
22:58:27<sandro>Anyone care to guess (or have any hard data) about the relative prevalence of escaped-markup vs rdf:XMLLiteral for using markup inside RDF?
23:06:03<oshani_>I was wondering what the minimum set of attributes (attributes as in RDFa) that a document should have to qualify it as a RDFa document?
23:42:19<Anchakor>akozak: I consider machine readability a hidden layer of content - every layer of content that seems as hidden to humans would probably be not meant for humans
23:45:27<akozak>Anchakor: but the semantic web is much more than just an extra layer of annotations on webpages
23:46:25<akozak>and does more than just provide more content
23:47:14<Anchakor>J:I think the comment fails to notice that presentation and data are still connected (whether using the semantic web approach or extracting content approch which mozilla jatpack does), but the publisher no longer dictates how the data are presented to user, but merely reccomends
23:47:59<PovAddict>J:
23:48:59<PovAddict>J:the author of that comment eventually said "ok you sold me :)"
23:49:04<Anchakor>akozak: yes, it is much more which is awesome, I was just thinking of one part of it - the one which the comment relates to
23:49:19<akozak>Anchakor: ah, I thought you were implying that that comment was a good challenge to the semantic web :)
23:49:55<Anchakor>akozak: by no means ;)
23:51:43<Anchakor>PovAddict: I guess I should get an account on slashdot (noo so much time lost! :)), but this was really a too nice comment related to semweb to let it go unnoticed :)
23:59:53<timbl>cygri, re glossing over the difference between "named RDF graph" and "RDF document with a URL" ... cwm really can't cope with them being the same. Completely different objkects. One is a graph, the other is something you can look up which may or may not actually be RDF at all.

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