00:06:18 Logomachist has joined #swig 00:12:52 timbl has quit 00:12:58 apspoliveira_ has quit 00:17:05 hg_ has quit 00:19:11 karlcow has quit 00:22:41 cygri has quit 00:31:37 Phae has joined #swig 00:33:15 jsoltren has quit 00:34:17 nathany has quit 00:36:06 LeeF has joined #swig 00:58:05 Phae has quit 01:02:12 shepazutoo has joined #swig 01:07:26 nathany has joined #swig 01:18:06 chimezie has quit 01:18:08 LotR has quit 01:18:08 gsnedders has quit 01:18:08 CaptSolo has quit 01:18:08 simeoni has quit 01:18:10 shepazu has quit 01:18:22 sessi has joined #swig 01:18:22 LotR has joined #swig 01:18:22 gsnedders has joined #swig 01:18:22 CaptSolo has joined #swig 01:18:22 simeoni has joined #swig 01:18:23 CaptSolo has quit 01:18:27 CS_ has joined #swig 01:18:46 simeoni has quit 01:19:26 karlcow has joined #swig 01:24:02 jaresty has joined #swig 01:26:46 nathany has quit 01:31:55 Arnia has joined #swig 01:38:52 shellac has quit 01:39:12 harbulot has quit 01:47:17 melvster has quit 01:53:34 melvster has joined #swig 01:53:55 melvster has left #swig 02:03:05 chimezie has joined #swig 02:15:46 Tristan has quit 02:16:14 Tristan has joined #swig 02:16:20 allisterb has joined #swig 02:18:51 MacTed has joined #swig 02:26:05 coreyleong has joined #swig 02:43:43 justben has joined #swig 03:03:25 chimezie has quit 03:13:07 eikeon has quit 03:14:34 chimezie has joined #swig 03:17:24 eikeon has joined #swig 03:19:36 Tristan has quit 03:21:56 Tristan has joined #swig 03:22:50 Tristan has quit 03:23:17 Tristan has joined #swig 03:24:11 Tristan has quit 03:24:32 Tristan has joined #swig 03:25:26 Tristan has quit 03:26:45 Tristan has joined #swig 03:29:30 coreyleong has quit 03:33:13 Tristan has quit 03:33:29 karlcow has quit 03:33:33 Tristan has joined #swig 03:40:28 drewp has quit 03:41:51 drewp has joined #swig 03:53:17 nathany has joined #swig 04:02:57 justben has quit 04:04:15 logger has quit 04:06:12 karlcow has joined #swig 04:08:38 dulanov has joined #swig 04:09:08 activefx has joined #swig 04:29:14 chimezie_ has joined #swig 04:30:40 chimezie has quit 04:32:16 chimezie_ is now known as chimezie 04:37:13 mun has quit 04:38:12 mun has joined #swig 04:39:15 dmitry_ulanov has joined #swig 04:44:27 activefx has quit 04:47:55 dulanov has quit 04:53:07 MacTed has quit 04:54:26 jansc has joined #swig 04:57:48 jansc has quit 05:15:32 cbichis has joined #swig 05:16:09 cbichis has left #swig 05:30:49 twanj has quit 05:31:41 oshani has quit 05:57:59 mhausenblas has joined #swig 06:00:59 dmitry_ulanov has quit 06:01:28 good morning Web of Data 06:05:20 kwijibo has joined #swig 06:06:26 oh oh - kwijibo 06:06:39 hi 06:06:53 can't believe it, man what are you doin up that early? 06:06:56 early for u mhausenblas ? 06:07:02 :) 06:07:20 i'm still in austria, so it's only 7 fro me 06:07:46 don't you ever sleep man? 06:07:47 ah 06:08:05 yep. from around 11:00pm to 6:00am 06:08:18 which is more than sufficient 06:08:36 (having children one learns to appreciate every hour sleep ;) 06:09:02 logger, pointer 06:09:17 oh, to early for the bots? 06:10:03 keithA has joined #swig 06:10:26 kwijibo has quit 06:10:35 logger-sioc-rdfa, pointer 06:10:35 See http://buzzword.org.uk/2009/logger-sioc-rdfa/swig/2009-03-11.html#T06-10-35 06:11:11 melvster, re http://esw.w3.org/topic/PushBackDataToLegacySourcesAuthentication - great work 06:12:18 keithA has quit 06:12:45 What I don't get is, though, the distinction between the RDF source and the non RDF-source 06:13:38 Our setup is: we have is a write-wrapper around a non-RDF-source such as Twitter 06:14:09 precisely as uriburner can be understood as a read-wrapper around Twitter 06:15:08 now, the user navigating in the 'semantic space' (i.e., pulling, viewing, filtering *RDF* data from different RDF data sources) want to edit 06:15:54 currently, the poor chap needs to copy & paste from Tabulator or whatever he uses, log into the origin non-RDF source and change things there 06:16:02 and pushback will shortcut this 06:16:16 where is the RDF source is in this setup? :) 06:44:02 so, we could add some stuff about how we gonna handle this in pushback as well (sort of plug-in?) 06:44:18 the RDForm is configurable per se 06:44:37 we can specify a default auth mechanism, people may choose others 06:45:09 today we will have a RDForm container that helps understanding the interaction part (Aftab is on this) 06:45:58 last things: I added a link to your auth page from the home page of pushback 06:47:24 phenny, tell melvster re pushback auth - thanks! and see http://buzzword.org.uk/2009/logger-sioc-rdfa/swig/2009-03-11.html#T06-10-35 for comments 06:47:24 mhausenblas: I'll pass that on when melvster is around. 06:55:29 dajobe: chump bot not around? 07:07:22 cbichis has joined #swig 07:09:32 sYskk has quit 07:11:52 cbichis has left #swig 07:14:53 presbrey has quit 07:20:45 jansc has joined #swig 07:28:22 kwijibo has joined #swig 07:33:53 Wikier has joined #swig 07:34:57 pauld has joined #swig 07:37:18 bengee has joined #swig 07:39:17 pesla has joined #swig 07:44:27 kwijibo has quit 08:02:54 timbl_ has joined #swig 08:04:32 Logomachist has left #swig 08:05:11 pauld has quit 08:10:46 Paul_Miller has joined #swig 08:12:45 tobyink has joined #swig 08:13:17 tobyink has left #swig 08:17:01 FabGandon has joined #swig 08:17:08 gromgull has joined #swig 08:18:49 nathany has quit 08:19:42 reto has joined #swig 08:24:45 IvanHerman has joined #swig 08:33:15 leobard has joined #swig 08:37:20 hg has joined #swig 08:46:47 duryodhan has joined #swig 08:47:38 reto has quit 08:48:03 hi .. total noob to semantic web/rdf etc. .... I wanted to use dbpedia .. any ideas/help ... for e.g, I have a list of articles about whom I want the infobox content -- using dbpedia's infobox crawl how do I get it ? 08:49:33 mmmmmrob has joined #swig 08:55:52 reto has joined #swig 08:59:59 reto has quit 09:05:12 you mean, how to get from wikipedia URLs to the DBPedia ones? 09:05:26 stetho has joined #swig 09:06:19 tobyink has joined #swig 09:08:36 duryodhan: basically, if you have a URL http://en.wikipedia.org/wiki/Resource_Description_Framework then the URI of the abstract thing which is RDF is http://dbpedia.org/resource/Resource_Description_Framework will give you a description of it 09:08:53 pauld has joined #swig 09:09:23 so if you GET that URI, you will first get a 303 redirect which basically means "actually, that's an abstract thing, but this is the description of it" 09:09:29 yeah but I have a huge list for which I need to get data .. so I downloaded the infobox dump directly 09:09:37 so I don't want to use the web 09:09:43 oh, ok 09:10:00 * KjetilK hasn't worked with the infobox dump 09:11:03 let me give an example .. 09:11:28 lets say I want to get the headquarter city and the url of all companies listed on nasdaq 09:11:38 to get their wikipedia pages is easy .. http://en.wikipedia.org/wiki/Category:Companies_listed_on_NASDAQ 09:12:18 but to get the city (or revenue or URL or whatever ) from the infobox on each page is a pain to do for me ... 09:12:25 so I thought I would use the infobox dump 09:12:31 or is there a simpler way to do it ? 09:13:05 now what I can do is for each wikipedia page I want info about, I do a grep in the dump and use the output ... but I am _sure_ there is a better way 09:13:21 libby has joined #swig 09:13:28 hmmm, yeah 09:14:07 shellac has joined #swig 09:14:26 hmpf, things started to get slow for me here, let me see 09:14:45 iand has joined #swig 09:15:25 kjetilk ? 09:15:28 * KjetilK needs newer hardware :-) 09:15:36 reto has joined #swig 09:15:43 aah .. I have lots of hardware .. what do you want to do ?? :D 09:16:56 basically, I think a SPARQL query could get you what you want, I just don't remember the properties you need to look for 09:17:12 * KjetilK kills firefox and starts over 09:17:50 what are properties ? dbpprop:website ? 09:18:00 right 09:19:11 so, you should be able to say stuff like SELECT * WHERE { ?company dbpprop:website ?website ; dbo:category } 09:20:52 that gave an error. . I changed it to dbpedia:category but that gave nothing as output too 09:21:19 yeah, that was just an example off my head without checking 09:22:01 yeah .. but whats wrong with SELECT * WHERE { ?company dbpprop:website ?website ; dbpedia:category } ? 09:22:38 hmmmm, so actually, there is a bit of reasoning involved, since the company isn't necessarily annotated with that it is listed on NASDAQ 09:24:02 eh ? 09:24:03 lets see if there is an easier way to do it 09:24:16 timbl_ has quit 09:24:22 basically, you need the YAGO ontology to find the stuff on DBPedia 09:24:36 I think skos:subject dbpedia:category Companies list on NASDAQ ? 09:25:44 right 09:25:53 you know the dataset better than me :-) 09:25:53 leobard has quit 09:26:00 but how do I search for that ? 09:26:05 so this is the page with it http://dbpedia.org/class/yago/CompaniesListedOnNASDAQ 09:26:07 I don't know SPARQL 09:26:43 yeah 09:28:27 but say now I want to do it for another category .. ? 09:28:35 all don't seem to have yago pages .. 09:28:45 if they have a mention in the ontology ... for e.g http://dbpedia.org/ontology/EducationalInstitution 09:28:51 how do I work with that ? 09:29:32 so if I want urls of all educationalInstitutions ? 09:30:04 duryodhan: http://sparql.pastebin.com/m74ac21fc 09:30:15 Run it at http://dbpedia.org/sparql 09:30:53 it is good to have people like tobyink around who actually knows the dataset :-) 09:31:33 * KjetilK got confused by the first example he found wasn't actually annotated with the NASDAQ URI, the data isn't perfect 09:31:41 I only really started looking at it properly a couple of weeks ago when I was trying to find a way of listing cheeses. 09:31:51 :-) 09:32:16 Generally I find the best way of querying it is to use the presence of infoboxes as a proxy to infer rdf:type. 09:32:18 we've been using some infobox data at work, but I didn't do the actual work 09:33:05 and we should probably have downloaded the infobox dump rather than use SPARQL, but I wasn't aware of the infobox dump before now 09:33:38 e.g. instead of looking for an rdf:type of , I would look for 09:34:35 tobyink: but then you are assuming that all cheese pages use infobox_cheese and not some other infobox 09:34:44 They do. 09:35:00 Or at least, they do more reliably than any other way of classifying them. 09:35:08 and if they don't, there isn't much you can do about it 09:35:29 It's a big, visible part of the page, so editors tend to notice its presence/absence. 09:35:53 the data isn't better than the effort gone into maintaining Wikipedia, but then Wikipedia is pretty good 09:36:21 * KjetilK trots off to focus on $work 09:37:57 tobyink: yeah .. but what I would do is start off from some page like .. Category: Type of Cheese 09:38:02 and then go to all the pages in it 09:38:23 http://en.wikipedia.org/wiki/Category:Cheeses 09:38:24 A: http://en.wikipedia.org/wiki/Category:Cheeses from duryodhan 09:38:43 start with that .. get all the names of pages .. and search in the infobox dump for them 09:39:24 * duryodhan was not only ignoring work .. he forgot about a meeting 09:39:36 got to run ... but please ping any thing you can think of 09:40:27 :-) 09:43:21 jaimico has joined #swig 09:45:25 duryodhan: URLs of universities - http://sparql.pastebin.com/m5c438e6c 09:49:41 mischat has joined #swig 09:51:21 cerealtom has joined #swig 09:51:29 cerealtom has left #swig 09:51:55 mmmmmrob has quit 09:56:27 swh has joined #swig 09:58:32 ldodds has joined #swig 10:02:00 BenO has joined #swig 10:03:17 jaimico has quit 10:07:39 Arnia has quit 10:14:33 ephemerian has joined #swig 10:21:17 tobyink: thats great .. it will help a lot .. but if I already have the name of the wikipedia page .. can't I do something with the infobox dump directly (cos from my experience I know that not all pages have infobox_university .. some may have infobox_name_of_univ 10:21:41 just a nagging concern .. what you gave will cover a lot of pages no doubt 10:21:53 Sorry, don't really know what an infobox dump is ?? 10:23:07 besbes has joined #swig 10:23:33 Arnia has joined #swig 10:24:02 http://downloads.dbpedia.org/preview.php?file=3.2_sl_en_sl_infobox_en.nt.bz2 10:24:03 B: http://downloads.dbpedia.org/preview.php?file=3.2_sl_en_sl_infobox_en.nt.bz2 from duryodhan 10:24:05 http://wiki.dbpedia.org/Downloads32 10:24:06 C: http://wiki.dbpedia.org/Downloads32 from duryodhan 10:28:19 duryodhan the bot picks up any urls with no space before them on the line (which is fine if you want it to! goes to swig.xmlhack.com) 10:28:30 ohh 10:28:31 sory it's a bit of a pain, but a space before it'll stop it 10:28:37 ok 10:28:39 thanks 10:28:46 :-) 10:29:32 jansc_ has joined #swig 10:29:48 OK, over 200 MB, not downloading that! But from a sample, it looks like the SPARQL query I provided will mostly work. Probably want to replace foaf:name with dbp:name and foaf:page with dbp:url. 10:29:49 jansc_ has quit 10:32:01 SPARQL can be run on the infobox dump locally using something like roqet . 10:32:43 ohh .. can you walk me through the query you provided ? I am totally new to SPARQL ... 10:34:04 Arnia_ has joined #swig 10:35:09 Arnia has quit 10:35:11 Arnia_ is now known as Arnia 10:35:49 Are you familiar with SQL though? 10:36:56 emrojo has joined #swig 10:37:02 sYskk has joined #swig 10:38:29 Let's assume you have roqet installed, and your infobox dump is saved with filename "infobox.nt". 10:39:12 Then a SPARQL query can be run on that file using the following command-line (I'm assuming here a familiarity with the Unix command line): 10:39:55 roqet -i sparql -r json -s 'infobox.nt' -e 'SPARQL GOES HERE' 10:42:07 yeah 10:42:14 I am asking about the SPARQL itself 10:42:27 minmax has joined #swig 10:42:29 why did you write the query the way you wrote .. what does that ?edu mean 10:42:30 We want to find a list of resources "foo" such that the following N-triples statement is correct : foo 10:42:52 So we use a placeholder/variable called "?foo" 10:43:14 SPARQL placeholders are named ?something or $something. The two are interchangeable. 10:43:37 SELECT * WHERE { ?foo . } 10:43:46 ok that I get now ... 10:43:50 whats the optional ? 10:43:54 Or we could be more specific: SELECT ?foo WHERE { ?foo . } 10:44:14 that would just give me the name of the page right .. nothing elsE ? 10:44:28 OPTIONAL is not supported by roqet (complain to dajobe). It's essentially analagous to a SQL LEFT JOIN. 10:44:40 It would give you the URI of the resource. 10:45:04 So, say we also wanted to get album cover art... 10:45:12 gromgull has quit 10:45:14 forget roqet for now .. just say the dbpedia.org/sparql page 10:45:23 ok 10:45:50 and how do I say to only get me results for USA ? 10:45:55 SELECT ?resource ?art WHERE { ?resource . ?resource ?art . } 10:46:37 The bit in the WHERE {} clause is essentially a template in Turtle which we want to be true for all results. 10:46:51 jansc has quit 10:46:51 and the . ? 10:47:06 The "?" just says "this is a variable". 10:47:30 However, for some albums, perhaps Wikipedia doesn't have the cover art, so no rows would be returned in the output at all. 10:47:40 gromgull has joined #swig 10:48:00 And that's where OPTIONAL comes in - you're saying that you still want to know about the album, even if it has no cover art. 10:48:14 SELECT ?resource ?art WHERE { ?resource . OPTIONAL { ?resource ?art . } } 10:48:48 no no .. the dot .. whats the dot for ? 10:48:51 or is that a new line ? 10:49:08 Similar to a SQL left join, where you're saying that you still want rows for the left table even if there is no correponding data in the right table. 10:49:15 dot is part of turtle. 10:49:36 It marks the end of a triple. 10:49:46 ok ok 10:49:56 and how do I say ... only get me results for Country : USA ? 10:50:01 (Though if you've only got one triple in a clause, you can leave it out.) 10:50:21 I would need something like "templateHasCountry" as one of the triple ? 10:52:05 LotR has quit 10:52:30 Something like this... http://sparql.pastebin.com/m7518cf92 10:53:00 LotR has joined #swig 10:53:49 but in that query whats the point of the optional part .. if you have done select * .. wouldn't it return everything it has for that entity ? 10:54:29 duryodhan: the * means: every variable in the query 10:54:38 ldodds: exactly. 10:54:42 not every property in the database 10:55:09 A triple store doesn't have a data dictionary like a relational database, so you have to match everything you're interested in, in the query pattern, in order to select it 10:55:53 ok ok 10:56:14 and ?edu dbpprop:country http://dbpedia.org/resource/United_States would also restrict to USA ? 10:56:39 * KjetilK had a good moment yesterday when my colleague said "actually, now I suddenly find SPARQL easier to write than SQL" :-) 10:57:21 KjetilK: cool :) 10:59:29 duryodhan: yeah, I would think so 10:59:36 duryodhan: SPARQL has "DESCRIBE" to do what you want "SELECT *" to do. 10:59:54 DESCRIBE's behaviour is a little implementation-specific though. 11:00:12 This works on dbpedia.org/sparql and returns a very big document - DESCRIBE ?edu WHERE { ?edu . ?edu . } 11:00:13 * KjetilK likes DESCRIBE a lot, it basically says "give me all you know about in RDF" 11:00:47 I hope we can specify DESCRIBE a bit more in the new SPARQL WG 11:01:33 KjetilK: argh, that reminds me that I forgot one of my suggestions to WG 11:01:48 yvesr: r u there ? 11:01:48 aklassen has joined #swig 11:01:55 I'll send it in anyway, hope the deadline wasn't too strict :) 11:02:28 * ldodds wants a DESCRIBE ... USING ... type feature to specify the algorithm 11:02:39 hmm so theoretically .. my original use case I could do DESCRIBE ?edu WHERE { ?edu foaf:page } ? 11:02:48 for each url I know of 11:02:59 ldodds: ah, yeah, like virtuoso does a "define" in the beginning? 11:03:28 not looked, but if it allows you to say I want a bnode closure, or other form of bounding the graph, then yes 11:03:38 ah, yup 11:03:42 duryodhan: Assuming that is a Wikipedia page, then yes, that should work. 11:04:29 ok great .. I won't do it .. but I think I am getting a hang of this .. 11:05:35 Good, good. The SPARQL spec actually has quite a lot of nice examples though none of them use dbpedia-specific properties, resources, etc. 11:06:07 duryodhan: some background on difference between the various SPARQL query forms (SELECT, DESCRIBE, etc): http://www.slideshare.net/ldodds/sparql-query-forms 11:06:09 phenny, tell yvesr : in the bbc music rdf I think you are using foaf:maker incorrectly, for example here : http://www.bbc.co.uk/music/artists/cc197bad-dc9c-440d-a5b5-d52ba2e14234, <#artist> foaf:made _:bnode, seems like the correct property to use (when talking about records). foaf:maker's domain is an owl:Thing and its range is a foaf:agent, whereas foaf:made is the other way round. 11:06:09 mischat: I'll pass that on when yvesr is around. 11:06:53 wow that describe is taking a lot of time on virtuoso .. maybe I shouldn't have run it on that ? 11:07:03 I just wanted to see some sample outputs 11:07:17 cygri has joined #swig 11:07:59 DESCRIBE will often return a *lot* of data. 11:08:15 There is no way of specifying how much you want. 11:08:37 Which is why it's usually best to use SELECT and a few fields that you're interested in. 11:10:24 chimezie_ has joined #swig 11:11:01 chimezie has quit 11:11:33 right now to get teh website I am doing ?edu dbpprop:website ?website . .. but this is on virtuoso 11:11:49 chimezie_ is now known as chimezie 11:11:52 will this work offline .. or will I need to use some URI instead of dbpprop:website ? 11:12:00 phenny, tell yvesr : I agree with mischat. Should use foaf:made, or perhaps dc:contributor 11:12:01 tobyink: I'll pass that on when yvesr is around. 11:12:05 karlcow has quit 11:12:24 foaf > dc 11:12:33 but yeah i agree with you too tobyink ;) 11:12:43 you could use dc terms namespace 11:12:48 Arnia has quit 11:12:54 the old dc namespace was quite messy re best practice for dc:creator etc 11:13:02 dc terms it is a bit more ontology-style 11:13:32 dbpprop:website relies on virtuoso knowing what "dbpprop:" is a shorthand for. I always tend to use full URIs in SPARQL as defining prefixes doesn't tend to save many bytes. 11:13:56 I always use xmlns:dc="http://purl.org/dc/terms/" 11:14:15 So when I say dc:contributor, I mean http://purl.org/dc/terms/contributor 11:15:51 which implies dc elements 1.1 contributor 11:16:10 ok 11:16:32 ah the joys of URIs 11:16:58 justben has joined #swig 11:18:23 phenny, tell yvesr Also, why http://www.perceive.net/schemas/relationship/ ?? The latest version is at this URI - http://purl.org/vocab/relationship/ - and has been for over 5 years 11:18:24 tobyink: I'll pass that on when yvesr is around. 11:19:17 ah yves is going to love us now ... 11:20:09 the relationship vocab doesn't have a nemesis relationship shame really 11:20:40 mischat : http://buzzword.org.uk/rdf/xen 11:21:03 nice 11:21:37 iand: http://purl.org/vocab/relationship/ uses owl:equivalentClass when it means owl:equivalentProperty 11:25:04 mischat : 11:25:21