Talk:LinkedData

Revision as of 13:35, 28 January 2009 by 140.147.245.139 (Talk)

Revision as of 13:35, 28 January 2009 by 140.147.245.139 (Talk)

January 27th #code4lib planning meeting


(10:12:04 AM) edsu: ok, so thanks for showing up, i guess there's only a few weeks till this thing goes down
(10:12:41 AM) edsu: the preconf as proposed can be found here: http://wiki.code4lib.org/index.php/LinkedData
(10:13:22 AM) edsu: i basically cut n' pasted from tomheath's iswc tutorial to see if it would get through
(10:13:40 AM) edsu: and last i heard there were 50 or so people signed up
(10:13:56 AM) rsinger: edsu: so i guess it worked :)
(10:14:27 AM) edsu: rsinger: yeah, heheh i think this whole thing was your and mike's idea originally :)
(10:15:15 AM) mjgiarlo: edsu: this is what you cribbed from? http://events.linkeddata.org/iswc2008tutorial/
(10:15:42 AM) rsinger: so what's our plan?
(10:15:44 AM) edsu: yes indeedy, i just added a link to some linked-data presentations at the end of the wiki page
(10:16:19 AM) edsu: well maybe we can all just take a second to review the Program Outline and see if it makes rough sense
(10:16:44 AM) edsu: http://wiki.code4lib.org/index.php/LinkedData#Program_Outline
(10:17:06 AM) edsu: also, if you have a demo idea, or don't agree with one I put in (just for brainstorming) plz change the doc
(10:17:09 AM) charper: First thought is that it's billed as two different pre-confs, w/ diff registration counts.
(10:17:16 AM) charper: Do we need a clearer division there?
(10:17:59 AM) rsinger: like morning == intro/'lectures', afternoon == 'breakouts'?
(10:18:11 AM) ***edsu looks at http://code4lib.org/node/266
(10:18:28 AM) rsinger: the program seems a little ambitious, at the moment
(10:18:35 AM) edsu: rsinger: agreed
(10:19:00 AM) edsu: yeah it seems to be billed as instruction in the AM, and collaboration in the PM
(10:19:05 AM) mjgiarlo: Insofar as possible, and this is highly general, but I would like to avoid talking about the Semantic Web, RDF, and the like. The more we can focus on the practical, the better. RDF seems somewhat inevitable but to some extent we can treat it as Yet Another Format?
(10:19:16 AM) lbjay: I think the outline looks pretty good but is light on the Practical Tasks
(10:19:42 AM) charper: Maybe we lead off the afternoon bits by working with folks on installs?
(10:19:46 AM) rsinger: meh, i think RDF should stay -- i would cut SPARQL
(10:19:50 AM) charper: Jena other triple store stuff.
(10:19:55 AM) lbjay: for instance, i think each item under "Publishing Linked Data on the Web" should have a Learn By Doing aspect
(10:20:54 AM) brocadedarkness: i agree, "learn by doing", focus on the easy and fun parts and try to actually get people to publish something
(10:20:54 AM) lbjay: but that could be left to the breakout sessions
(10:20:58 AM) mjgiarlo: edsu and I (and maybe others on IRC) talked about encouraging folks to bring a dataset or have some data in mind to try to model via linked data. and those who don't, well, can do FOAFs?
(10:21:16 AM) charper: Bringing data is good.
(10:21:21 AM) lbjay: that way folks could segregate based on language preference
(10:21:21 AM) brocadedarkness: sounds good
(10:21:27 AM) iand: rsinger: I agree, sparql could be cut
(10:21:29 AM) charper: But if not, even without foaf stuff, why not show people how to pull data.
(10:21:39 AM) charper: See - that's where I thought sparql=good...
(10:21:52 AM) rsinger: i just think sparql is a day to itself
(10:21:52 AM) charper: Need a data set - run some queries against dbpedia...
(10:22:04 AM) iand: I think the focus should be on publishing
(10:22:17 AM) iand: otherwise the day is going to be too crammed
(10:22:19 AM) mjgiarlo: rsinger: I think it sort of has to stay, but I wouldn't want to say much more than "hey, RDF is this model that has these things called triples that neatly map to natural language subject, predicate, object." Fair? Too simple?
(10:22:25 AM) rsinger: a mention of sparql, maybe an example
(10:22:48 AM) lbjay: i was thinking more "here's how to create a sparql endpoint" vs. a full sparql tutorial
(10:22:54 AM) iand: mjgiarlo: probably slightly too simple, but I wouldn't go into a huge amount of depth
(10:23:03 AM) edsu: lbjay: you could definitely imagine that as a breakout session in the afternoon
(10:24:01 AM) edsu: iand: how could we focus more on the publishing side?
(10:24:46 AM) edsu: mjgiarlo: i also am a bit concerned that simply explaining what rdf is could take a while for some
(10:24:49 AM) iand: edsu: I think it mostly is, just that introducing sparql beyond "here it is, this is what it does" could take a long time
(10:24:50 AM) mmmmmrob: edsu: maybe a session on modelling data, looking at ontologies?
(10:24:54 AM) dchud: here's a thing that would've made the solr preconf better - if somebody had a pre-packaged instance they could share around
(10:25:35 AM) rsinger: if only there were some accessible triple store in the cloud...
(10:25:38 AM) dchud: some 5-50Mb of data sitting next to a basic server so anybody there could have something running locally to poke at
(10:25:41 AM) rsinger: if only...
(10:25:46 AM) dchud: and we should assume the network will fail us
(10:25:49 AM) janusman: dchud: instance for what application?
(10:25:51 AM) rsinger: :)
(10:26:03 AM) iand: edsu: I did an intro to rdf as part of a linked data presentation tom heath and I gave to the guardian newspaper... took about 10 minutes to a mixed tech/non-tech audience
(10:26:09 AM) janusman: dchud: you mean besides example/ ? =)
(10:26:14 AM) mjgiarlo: the eight-hundred pound platform in the room.
(10:26:21 AM) dchud: janusman: doesn't matter. some basic "take this sqlite file and run a web server serving the data in linked style" in a little bit of code
(10:26:34 AM) edsu: iand: that sounds better than trying to wade through cereal tom's slides
(10:26:45 AM) dchud: janusman: sorry, what example do you mean?
(10:26:54 AM) edsu: iand: no offense to those slides, they are great, but the iswc audience won't be the same i don't think
(10:26:57 AM) rsinger: i feel like an intro to rdf wouldn't be bad
(10:27:01 AM) janusman: dchud: oh just the example/ directory packaged with solr =)
(10:27:06 AM) mjgiarlo: 10 minutes sounds like the sweet spot to me
(10:27:17 AM) rsinger: iand's intro to rdf presentation was the 'aha' moment for me
(10:27:25 AM) iand: rsinger: what about a busy developer's intro to rdf... in 10 mins
(10:27:33 AM) iand: I'm happy to do that
(10:27:44 AM) dchud: janusman: yeah something like that would be good, but nobody had that pre-packaged for people there that day, iirc
(10:27:45 AM) mmmmmrob: that helped me
(10:27:55 AM) janusman: I don't quite get the "linked data" rush. I perhaps need some evangelism =)
(10:28:21 AM) dchud: erikhatcher had prepped some data, but only a few people had downloaded it before the event, so when the net went wonky, many were stuck without it
(10:28:27 AM) janusman: dchud: ah =)
(10:28:29 AM) edsu: janusman: it's not a rush it's been taking 8 years so far :)
(10:28:33 AM) rsinger: let me add that the 'aha' moment was pre-talis :)
(10:28:41 AM) iand: heh,
(10:29:26 AM) ***mmmmmrob arrived late - is there a draft agenda for the day anywhere?
(10:29:28 AM) lbjay: +1 on the pre-packaged instance idea
(10:29:35 AM) lbjay: mmmmmrob: http://wiki.code4lib.org/index.php/LinkedData
(10:29:39 AM) mjgiarlo: rsinger: didn't you already earn your pay today with your "if only" comment, shillboy?
(10:29:45 AM) mmmmmrob: lbjay++
(10:29:48 AM) ***edsu is axing the program outline
(10:30:07 AM) edsu: iand: does your intro to RDF cover linked-data-ness?
(10:30:21 AM) mmmmmrob: edsu: that lcsh.info demo will be quick...
(10:30:30 AM) edsu: mmmmmrob++
(10:30:49 AM) rsinger: so i'm not sure how the prepackaged demo would work
(10:31:03 AM) iand: edsu: I'll makes sure it does... current one needs a bit of updating
(10:31:08 AM) rsinger: BenO wrote recently about the travails of getting a triple store running locally
(10:31:16 AM) dchud: rsinger: a little data. a little code. "take this, copy it, and run it."
(10:31:18 AM) edsu: /reload http://wiki.code4lib.org/index.php/LinkedData#Program_Outline
(10:31:25 AM) jtgorman: I'm not fully paying attention, but if there's desire to show how someone can take a dataset and make it "linked" you might want to look into that dataset of circulation information that was released a while ago. Just started listening to a Talis podcast that had a little more info this morning
(10:31:41 AM) edsu: jtgorman: +1 to that
(10:31:52 AM) ***jtgorman goes back to the day job
(10:32:04 AM) mmmmmrob: jtgorman: dave patten at huddersfield
(10:32:07 AM) janusman: so... let's say I have a Solr search inside a Library Catalog. Should I then code RDF into the facets for "authors" so that something can understand a search for "bananas" has to do with a certain author...? or what would be a good example of linked data and Solr?
(10:32:36 AM) edsu: mmmmmrob: yeah, he released the data then this guy patrick tried to convert it to linked-data
(10:32:42 AM) janusman: or just that the author facet is indeed a "person" who can then be looked up, etc?
(10:32:49 AM) edsu: mmmmmrob: i think he may be going to code4lib now that i think of it
(10:32:53 AM) iand: I think some practical examples of publishing library-oriented data as linked data need to be on the agenda
(10:33:03 AM) charper: I'm still stuck on whether the focus should be entirely publishing.
(10:33:03 AM) anarchivist: iand: agreed
(10:33:06 AM) mmmmmrob: if only...
(10:33:07 AM) rsinger: agreed
(10:33:22 AM) dchud: iand: agree
(10:33:24 AM) charper: Publishing makes sense, but it doesn't get at *why*
(10:33:33 AM) rsinger: charper: well, publishing and consuming what's published, right?
(10:33:34 AM) charper: Don't we want to be talking about using this stuff?
(10:33:46 AM) charper: rsinger: exactly my point
(10:33:50 AM) dchud: charper: agree. i'd be up for a brainstorm session on practical apps and some hack-fun
(10:34:01 AM) ***janusman also would like to know *why* and maybe he'll get excited
(10:34:09 AM) rsinger: charper: but you don't need sparql for that, do you?
(10:34:29 AM) charper: rsinger: no you don't, but it helps, right?
(10:34:37 AM) lbjay: janusman: we're trying to hash out an agenda for a preconf. we are beyond the why at this point
(10:34:57 AM) janusman: lbjay: Ok, will ask the intertoobz then ;)
(10:34:58 AM) charper: example being the stuff Martin M did, with querying his author file against dbpedia's sparql endpoint.
(10:35:01 AM) rsinger: charper: right, but do you have a way of adding it in, say, 30 minutes?
(10:35:01 AM) charper: Creating links.
(10:35:27 AM) charper: rsinger: true. Probably too ambitious.
(10:35:27 AM) lbjay: not that the preconf shouldn't include some "why?" aspects
(10:35:36 AM) edsu: it definitely should imho
(10:35:42 AM) rsinger: me too
(10:36:54 AM) lbjay: wherefore art thou, RDF?
(10:36:57 AM) edsu: pkeane: hey :)
(10:37:06 AM) pkeane: greetings
(10:37:30 AM) edsu: pkeane: you aren't going to code4lib in providence by any chance are you?
(10:37:32 AM) mjgiarlo: janusman: If you google around for "library sweden linked data", you can see how linked data can work in the OPAC.
(10:37:44 AM) rsinger: charper: right, i mean, i think it makes sense for martin m to show off libris and explain that
(10:37:54 AM) pkeane: edsu: unfortunately not (wish I was)
(10:37:59 AM) janusman: mjgiarlo: thanks!
(10:38:05 AM) edsu: yes, just added libris.kb.se to list of possible demos
(10:38:27 AM) edsu: jphipps: i'd also like to hear more about the registry and vocabulary development (RDA, etc)
(10:38:59 AM) mjgiarlo: So re: the program, we ought to have a section on what linked data is and what its value is directly before or after iand's intro to RDF?
(10:39:03 AM) ***iand on phone
(10:39:18 AM) jphipps: can do, but maybe charper should talk about RDA
(10:39:31 AM) mmmmmrob: RDFa ?
(10:39:36 AM) edsu: so do you all think the AM could be: intro to rdf/linked data (iand) followed by various demos/presentations to get people seeig various applications?
(10:39:58 AM) mjgiarlo: +1 more or less
(10:40:00 AM) rsinger: charper: but without the critical mass of published data, it's all moot anyway, right?
(10:40:04 AM) jphipps: edsu: yes
(10:40:17 AM) edsu: then in the afternoon people could aggregate around the projects/demos that interested them?
(10:40:31 AM) edsu: also leave some space for other people in the audience who might want to present in the AM?
(10:40:31 AM) rsinger: this sounds like a good plan to me
(10:40:39 AM) charper: rsinger: yes, it all does become moot - though there's a critical mass forming outside of libs.
(10:40:40 AM) edsu: unconference style?
(10:40:51 AM) mjgiarlo: edsu: yeah, and some of us can be floaters, helping folks who want to publish FOAF, etc.
(10:40:54 AM) lbjay: i'm skeptical that's enough tutorial
(10:41:00 AM) charper: jphipps: I'd be happy to talk a bit about the RDA work itself, though your just as qualified as me.
(10:41:06 AM) edsu: lbjay: yes, me too
(10:41:28 AM) jtgorman: one of my biggest regrets about missing the conference will be missing the chance to do this linked-data thing :(
(10:41:46 AM) ***jtgorman will be watching blog posts about it closely
(10:41:51 AM) lbjay: edsu: maybe not enough for folks to be capable of breaking out independently
(10:42:04 AM) rsinger: plus, i think it's imperative to explictly lay out some ideas of what and how LOD in libraries should look like
(10:42:09 AM) mjgiarlo: jtgorman: and I'm sure we'll have documentation, or maybe audio or video if that happens.
(10:42:14 AM) edsu: maybe we could follow up iand's talk with a practical exercise of creating a foaf file?
(10:42:34 AM) edsu: or maybe using bibont to describe a book/article?
(10:42:36 AM) rsinger: i'm afraid FOAF isn't going give them the 'aha'
(10:42:42 AM) rsinger: edsu: that might be better
(10:42:56 AM) edsu: to get people engaged with what makes it different than say just using mods or something?
(10:43:04 AM) dchud: the macro view of linked data's linkedness and HTTP issues interests me more than the micro issues of RDF and some RDF format examples
(10:43:08 AM) lbjay: i'd like to see an exercise on how to discover the vocablularies to define whatever dataset they're interested in
(10:43:34 AM) mjgiarlo: edsu: also, the stuff in your "follow your nose" post could be aha-inducing. it's a powerful metaphor.
(10:43:43 AM) lbjay: here's my data. how do i find some vocabularies and when do a need to define my own terms
(10:43:52 AM) lbjay: s/a/I/
(10:43:53 AM) charper: lbjay: that gets back to mmmmmrob's earlier point about data-modeling, too.
(10:43:55 AM) edsu: lbjay: nice, just added that under demos
(10:44:00 AM) rsinger: lbjay: and what are the ramifications of either?
(10:44:11 AM) lbjay: rsinger: i'd like to know that myself
(10:44:22 AM) rsinger: me too
(10:44:29 AM) edsu: i think it also would gell nicely with jphipps talking about registries, and maybe iand again talking about openvocab
(10:44:53 AM) edsu: lbjay: so this would get back to encouraging people to bring some data
(10:45:11 AM) charper: edsu: a good segue to the problem that there often isn't a vocab out there.
(10:45:21 AM) rsinger: dchud: i agree -- we need to keep the eye on the prize of 'LOD'
(10:45:40 AM) iand: I was thinking a bit of practical examples: describe a book, describe an article, describe a journal, describe a person, describe a library location
(10:46:38 AM) rsinger: describe a map
(10:46:54 AM) iand: edsu: I am happy to talk about open vocab, I also have paget which is an OSS PHP framework for publishing linked data - but it's quite alpha
(10:47:47 AM) lbjay: is there a simple linked data app that would run under something like google app engine?
(10:48:05 AM) iand: lbjay: I'm not aware of one
(10:48:17 AM) rsinger: dchud: the only advantage that i see focusing on rdf brings to the table is that shakes things up from a document-centric POV to a resource-centric POV
(10:48:27 AM) lbjay: something that folks could easily install locally. this is getting back to the pre-packaged examples idea
(10:48:35 AM) iand: rsinger: is it worth showing xulu?
(10:48:53 AM) iand: rsinger: it demonstrates the resource-centric nature of linked data pretty well
(10:49:04 AM) edsu: dchud: would you be interested in talking about http/html use in the am for potential breakout in the afternoon?
(10:49:22 AM) rsinger: iand: yeah, possibly. i guess it depends on whether or not robotrobot has completely broken it by then
(10:49:30 AM) mjgiarlo: I added a couple of the bullet points from cerealtom's program outline to the tutorial section of ours just to flesh things out.
(10:50:01 AM) edsu: so what would go into a pre-packaged example?
(10:50:11 AM) edsu: a little webserver serving up linked data?
(10:50:20 AM) lbjay: edsu: that is the half-baked idea
(10:50:32 AM) edsu: i kind of like the idea of using rdfa
(10:50:38 AM) edsu: for the pre-packaged part
(10:50:47 AM) rsinger: it'd be easier to distribute
(10:51:02 AM) mjgiarlo: edsu: is there a data set like lcsh but, uh, less political and could be used for the aha demo? lcsh.info was basically perfect for this and you've already built some python utils.
(10:51:25 AM) edsu: mjgiarlo: i could easily run lcsh.info off my laptop
(10:51:38 AM) iand: dare I mention that talis platform is basically pre-packaged linked data? to achieved balance in the universe I'll also mention that virtuoso can be installed on ec2
(10:51:38 AM) lbjay: the thesaurus thing... someone posted to lod-public recently
(10:51:50 AM) lbjay: t4gm.info
(10:52:02 AM) edsu: lbjay: that's what i was going to say
(10:52:13 AM) rsinger: iand: dchud raised a good point that network connectivity isn't a given
(10:52:42 AM) iand: rsinger: ok, that's important
(10:52:55 AM) mjgiarlo: edsu: yeah, I just wouldn't want any shit to come your way should the Powers That Be at LC get wind.
(10:53:51 AM) edsu: i don't even understand what a pre-packaged demo would do
(10:54:04 AM) rsinger: however, we could always have virtuoso (or something - YARS, Sesame) on a local, ad-hoc network
(10:54:14 AM) rsinger: me neither, really
(10:54:39 AM) rsinger: i feel the prepackaged demo is a little too prescriptive
(10:54:46 AM) lbjay: pre-packaged sandbox
(10:55:20 AM) iand: back on the practical examples, I'd like to take some pre-existing data and show how you can convert it to RDF for publication as LD
(10:55:21 AM) rsinger: esp. since it's feasible that all some people need to 'publish linked data' is xslt
(10:55:53 AM) iand: rsinger: true, the main point is understanding what vocabularies to use and how to model the data
(10:56:26 AM) ***rsinger nods.
(10:57:00 AM) charper: iand: does your gentle intro get into the modeling questions?
(10:57:06 AM) edsu: iand, rsinger that's the important point i think
(10:57:24 AM) ***edsu adds iand's practical example
(10:57:49 AM) iand: charper: no, it's just background to help understand how it all works
(10:58:02 AM) rsinger: i think it's important to mention the various ways that LOD can be achieved
(10:58:12 AM) edsu: iand: do you see that example of working through a data set as something we'd do in the AM?
(10:58:29 AM) lbjay: rsinger: as in RDFa, conneg, etc?
(10:58:36 AM) iand: on rsinger's point it'll be good tosay that you don't need to convert everything to a triple store, so long as you can template your output to RDF
(10:58:42 AM) rsinger: lbjay: yeah, exactly
(10:58:50 AM) rsinger: lbjay: the pragmatic approach to publishing
(10:59:06 AM) iand: edsu: I think it'll really help connect people to the concepts using data they are already familiar with
(10:59:09 AM) rsinger: lbjay: the, we can't all run a triplestore easily, approach
(10:59:13 AM) ***lbjay agrees
(10:59:45 AM) iand: edsu: i.e. assign URIs to important resources like people, places, things, events
(10:59:55 AM) charper: Other level of that is the various levels of utility to the LOD concept.
(11:00:10 AM) mjgiarlo: jbrinley: you have exhausted my java/jvm knowledge, sadly.
(11:00:20 AM) charper: Difference between what iand just said vs all the depth and specificity of things like the rda work and bibont and the like.
(11:00:29 AM) charper: diff levels of modeling, if you will.
(11:00:36 AM) janusman: LOD = Level of Detail?
(11:00:45 AM) mjgiarlo: Linking Open Data
(11:00:54 AM) rsinger: charper: and the advantage to just having URIs
(11:00:58 AM) mjgiarlo: AKA Linked Data On the Web
(11:01:10 AM) rsinger: charper: if that's all you can reasonably provide
(11:01:12 AM) lbjay: Lords Of [the] Dance
(11:01:28 AM) charper: LOL - an accidental acro-homonym.
(11:01:31 AM) rsinger: lbjay--
(11:02:22 AM) charper: rsing: so my question is, is that the focus of this pre-conf?
(11:02:25 AM) edsu: would it be confusing/heresy do you think to introduce the idea of low-barrier linking via <link> like autodiscovery in html?
(11:02:48 AM) rsinger: edsu: no
(11:02:59 AM) charper: gets back to dchud's point about avoiding the RDF bits altogether, but then you lose the resource-centric rather than record-centric nature of this space.
(11:03:13 AM) iand: edsu: heresy yes, but not confusing :)
(11:03:32 AM) rsinger: charper: i don't think i can answer that -- personally, i think the library world is at the 'basic introduction' stage
(11:03:39 AM) edsu: iand: :)
(11:03:49 AM) rsinger: charper: and i think this helps pave the path for RDA
(11:04:19 AM) rsinger: (which, obv. is not a decision that the technologists will have to make)
(11:05:05 AM) lbjay: edsu: +1 on low-barrier <link>-ing
(11:05:30 AM) rsinger: charper: but i think we're not even at the Kool-Aid stage, I don't think libraries even are aware there's a grocery store that stocks Kool-Aid
(11:05:41 AM) edsu: i'm worried if the bar gets set that low people won't make the effort to understand what rdf brings to the table
(11:06:01 AM) lbjay: edsu: that's what i read from rsinger's "all the various ways that LOD can be achieved"
(11:06:44 AM) charper: rsinger: I'm not entirely sure that's true. I get so many fewer blank stares when I talk about this stuff than a few years back.
(11:06:44 AM) rsinger: edsu: i think what rdf brings to the table either has to be self-evident, or the advantages need to be shown
(11:06:50 AM) edsu: rsinger: i'd love to see a brief history of the semweb, that ram the layer cake down people's throats
(11:07:00 AM) edsu: s/that ram/that doesn't ram/
(11:07:02 AM) charper: We had almost 200 people in a session on this stuff at ALA this weekend.
(11:07:03 AM) edsu: :)
(11:07:09 AM) pkeane: edsu: +1 would like to see <link> at least considered part of LOD
(11:07:13 AM) charper: edsu+-
(11:07:16 AM) edsu: charper: how did that go?
(11:07:23 AM) charper: Went really well.
(11:07:39 AM) charper: Got a little weird and ExLibris-y at the end, but that was to be expected...
(11:07:51 AM) rsinger: charper: but this was more from a cataloger's perspective than a technologist, though, right?
(11:07:52 AM) edsu: charper: because of an exlibris product?
(11:08:07 AM) lbjay: charper: was K Harnish in attendance?
(11:08:09 AM) charper: rsinger: yes, it was more cataloger based - that's very true.
(11:08:18 AM) charper: edsu: yes because of product
(11:08:20 AM) rsinger: charper: i think that's a much easier leap for them
(11:08:22 AM) dchud: charper: why do you need RDF to talk resource-centric? we're talking about linking web resources, right? they can be anything.
(11:08:32 AM) charper: lbjay - k harnish was co-presenter on panel with myself and Diane Hillman.
(11:08:45 AM) charper: dchud: that's very true
(11:09:03 AM) iand: edsu: +1 for no layer cake
(11:09:22 AM) rsinger: dchud: but majority of 'web resources' are 'documents'
(11:09:24 AM) dchud: isn't <link> kind of the whole point? we could spend a whole day on that. hey, let's spend the whole day on that!
(11:09:32 AM) charper: Maybe I'm thinking too much from the cataloger perspective, but I find that many people get stuck focusing on the record.
(11:09:39 AM) edsu: dchud: would you want to talk about that in the AM?
(11:09:45 AM) rsinger: exactly
(11:09:53 AM) edsu: dchud: for possible breakout in the afternoon?
(11:09:55 AM) rsinger: charper: er, that was to you :)
(11:09:56 AM) mjgiarlo: +1 - no layer cake, focus on low-hanging fruit/link-header stuff
(11:09:57 AM) iand: charper: I agree, and that's what documents are in the web of data: records
(11:10:02 AM) dchud: edsu: just point me at a mic and wind me up
(11:10:16 AM) brocadedarkness: charper: i wish people would look less at the record. you are completely right
(11:10:21 AM) lbjay: charper: and by "people" you mean ex libris? ;)
(11:10:46 AM) edsu: dchud: :) (adding to possible demos section)
(11:10:50 AM) dchud: if you spend two hours introing rdf and modeling, i will tune out and use that time to work on some kind of demo
(11:10:50 AM) charper: lbjay: no, no - not at all. I mean catalogers, though our vendor's certainly play into that...
(11:10:55 AM) brocadedarkness: maybe stuff for a different conference/day, but both RDA and RDF/Linked data are to me less about records and more about links
(11:11:15 AM) iand: I don't think the session should be called linked data if it's not talking about URIs + RDF to link em together, otherwise it's diluting a well defined term
(11:11:48 AM) iand: call it something else like "publishing data"
(11:11:51 AM) dchud: iand: maybe i should have checked my dogma at the door.
(11:12:12 AM) charper: iand: to play devil's advocate, could we have the focus be linked data, but let dchud make the point that the URI's as identifiers are useful without the RDF too?
(11:12:15 AM) iand: dchud: huh, I'm the one being dogmatic about LOD :)
(11:12:42 AM) mjgiarlo: I am allergic to dogma dander.
(11:12:43 AM) dchud: iand: i thought i was :)
(11:12:46 AM) edsu: iand: tbl's 4 rules of linked data don't say anything about rdf do they?
(11:13:00 AM) rsinger: i think charper is on the money
(11:13:07 AM) charper: sorta makes the point that consistent identification of resources is valuable whether you've drank the kool-aid or not.
(11:13:22 AM) iand: charper: yes they are useful, and they are the most important part of linked data, but the meaning of the relationships between the resources is almost as important
(11:13:33 AM) edsu: i think what dchud's talking about is more REST than it is linked-data
(11:13:40 AM) dchud: we're going to have a roomful of people with interesting data or data interests. if we can get 4-10 data sets up and online and linky by the early afternoon, we can then spend a few hours hacking on what we can do with that
(11:13:46 AM) dchud: all formats are welcome
(11:13:49 AM) charper: iand: I couldn't agree more, but I drank the kool-aid a good 6-7 years ago. ;)
(11:13:58 AM) dchud: edsu: no i mean the linking bits
(11:14:15 AM) iand: dchud: they won't get on the LOD cloud unless they conform to the conventions of URIs + RDF + links to other data sets
(11:14:22 AM) pmurray: robcaSSon: ping
(11:14:43 AM) iand: not that getting on the cloud diagram is the only possiblt outcome, but it looks great for the community
(11:14:49 AM) dchud: iand: i haven't drank enough punch to see yet why that would concern anybody :)
(11:15:04 AM) dchud: i'd like to walk away from this day-long session to understand why it might
(11:15:08 AM) edsu: i think it's important to build a community
(11:15:11 AM) dchud: i mean, understanding more about why it might
(11:15:12 AM) edsu: of people doing roughly the same thing
(11:15:22 AM) rsinger: i agree
(11:15:40 AM) edsu: but pretending like there isn't divergence of opinion isn't going to do anyone any favors
(11:15:54 AM) edsu: although it could be confusing :)
(11:16:06 AM) charper: iand: but could that come from 2 diff places? Once you've got the URI's, a 2nd-party can use those in RDF and get on the cloud.
(11:16:34 AM) iand: edsu: you're right about timbl's 4 rules, but in the sentence before them he says use RDF for data http://www.w3.org/DesignIssues/LinkedData.html
(11:16:40 AM) ***iand in pedant mode
(11:16:42 AM) charper: edsu: Show how the divergence of opinion is a good thing. dchud mints some uris, I can then do stuff with that whether or not he cares about the cloud.
(11:16:50 AM) rsinger: again, i think the convincing argument has to be made by either side
(11:17:04 AM) iand: charper: yes, publishing linkable identifiers is the most important
(11:17:05 AM) jphipps: it's more important to sow understanding than confusion
(11:17:30 AM) mjgiarlo: To get away from dogma for the moment (citing my previous allergy), at the very least, we ought to be able to answer: 1) Hey, what's this linked-data thing? 2) Well hey, that's awesome, but how do I take this pile of data and make it like the swedish library or lcsh.info site?
(11:17:34 AM) iand: rsinger: is the day going to be a rehash of old arguments or is it going to be a linked data day?
(11:17:43 AM) edsu: jphipps: yeah, this is true
(11:18:20 AM) jphipps: if we can't agree on the nature of _effective_ LOD, then no attendee will leave understanding what we're talking about or why it's important
(11:18:23 AM) rsinger: iand: to me, i would focus entirely on rdf
(11:18:30 AM) dchud: mjgiarlo: agree, and add 3) oh cool, and here's what i can do with it
(11:18:31 AM) charper: Maybe dogma and re-hashing old arguments could be a breakout session. :)
(11:18:50 AM) rsinger: but i guess my point is that if there's an 'aha' moment that needs to be had, let's not assume it's been there before they get in the room
(11:18:51 AM) dchud: heh maybe i'll make a dogma classification system
(11:19:28 AM) lbjay: mjgiarlo: exactly, and #2 is why i thought a pre-packaged sandbox would be useful.
(11:19:39 AM) mjgiarlo: rsinger++ # aha-ifying is kind of key to the AM session
(11:19:41 AM) lbjay: mjgiarlo: not that i've figured out what that would be yet
(11:20:05 AM) iand: do c4lib attendees need to be convinced of importance of identifiers? I hope not. but maybe they need to be convinced of utility of identifiers that can be put in a web browser and used to retrieve more info about the identified thing
(11:20:18 AM) rsinger: lbjay: isn't dbpedia kind of the 'prepackaged' sandbox?
(11:20:34 AM) lbjay: rsinger: i guess it could be
(11:20:53 AM) mjgiarlo: iand: it's hard to say in advance. the dynamic of the conf has shifted a bit, IMO. I mean, we have an OLE project update this year.
(11:21:00 AM) rsinger: i guess we need to throw the 'network connectivity optional' thing out the window because we're talking about data DATA ON THE WEB
(11:21:08 AM) ***mjgiarlo implies a "for fuck's sake"
(11:21:15 AM) rsinger: maybe even data Data DAta DATA on the web
(11:21:21 AM) edsu: rsinger: :) yea
(11:22:13 AM) rsinger: well, the OLE people really need to be shown the kool-aid, imo
(11:22:18 AM) rsinger: hopefully some will be in attendance
(11:22:38 AM) iand: ok, sounds like we need a really strong piece right up front to give audience first chance of the aha moment... then each demo after gives a better chance of the aha
(11:22:44 AM) mjgiarlo: rsinger: don't disagree at all.
(11:22:46 AM) pmurray: rsinger: Sorry -- just glancing over. What kool-aid are we supposed to be drinking?
(11:22:53 AM) edsu: iand: i like the sound of that
(11:22:58 AM) rsinger: pmurray: linked data
(11:22:59 AM) iand: then in PM, after they've all stopped weeping, we set the world right
(11:23:07 AM) pmurray: ("we" == "I'm participating in the OLE Project")
(11:23:11 AM) ***mbklein is really looking forward to receiving the aha.
(11:23:15 AM) rsinger: Ian Davis, layer of hands
(11:23:37 AM) mjgiarlo: "He dropped a platform on my head and I felt ALL BETTER!"
(11:23:38 AM) pmurray: rsinger: Not sure how that would be appropriate; OLE is not providing a user interface to the web
(11:23:56 AM) rsinger: pmurray: neither is linked data
(11:24:04 AM) dchud: gotta go, will check the wiki later
(11:24:09 AM) edsu: lets not get silly now
(11:24:13 AM) edsu: :)
(11:24:22 AM) iand: edsu: I could morph my rdf intro into a "why and what is linked data"
(11:24:26 AM) pmurray: rsinger: Okay. True. Let me rephrase.
(11:24:59 AM) pmurray: rsinger: OLE is about providing services. If the underlying services make use of linked data concepts, then all is well.
(11:25:08 AM) rsinger: pmurray: there we go
(11:25:09 AM) edsu: iand: i kinda liked the idea of just talking about rdf up front, and maybe I or someone else could do a linked data intro (the LOD cloud, etc)
(11:25:16 AM) pmurray: s/providing services/providing a framework for services/
(11:25:27 AM) edsu: iand: using your honed guardian talk i mean
(11:25:29 AM) rsinger: ...NOW THAT DCHUD HAS LEFT THE ROOM...
(11:25:34 AM) iand: edsu: cool
(11:25:38 AM) rsinger: let the rdf sweep over the agenda
(11:25:56 AM) edsu: i secretly suspect that people who are anti-rdf just don't understand it
(11:26:09 AM) edsu: which is a good enough reason i suppose
(11:26:11 AM) charper: edsu: we should serve kool-aid before lunch...
(11:26:32 AM) edsu: might work better if we all drop acid :)
(11:26:34 AM) rsinger: "Here is your juice box"
(11:26:47 AM) rsinger: "Poke straw through hole"
(11:26:48 AM) charper: I think the other bit of anti-rdf-ness is that it's XML serialization is possibly the ugliest thing in tech-land...
(11:26:53 AM) iand: edsu: I'm not anti quantum chromodynamics but...
(11:26:57 AM) mjgiarlo: "Is it me, or does this smell like rohypnol?"
(11:26:59 AM) rsinger: "Your data is now magically linked"
(11:27:03 AM) mjgiarlo: "No, no, that's RDF!"
(11:27:33 AM) edsu: iand: :)
(11:27:35 AM) iand: bring on the hypnotoad
(11:27:43 AM) rsinger: charper: but, remember, these people have to deal with MARCXML
(11:27:58 AM) anarchivist: acid++
(11:28:13 AM) edsu: ok, so please feel free to edit the wiki page if something is missing
(11:28:16 AM) iand: yeah I don't think awful serialisations should be offputting to the lib crowd
(11:28:23 AM) charper: rsinger: right, and I keep forgetting they aren't catalogers... My audience changes too much.
(11:28:23 AM) lbjay: mmmmmrob: are you coming to providence?
(11:28:32 AM) edsu: otherwise lets ruminate on this a bit more and connect up again in a few days, or early next week?
(11:28:38 AM) mjgiarlo: Think we ought to introduce the topic of linked data before launching into what is RDF, though. Even if it's just a quick intro.
(11:28:43 AM) rsinger: edsu: definitely
(11:28:49 AM) iand: mjgiarlo: yes I agree
(11:28:54 AM) rsinger: edsu++ # I think this is begin to make sense
(11:29:00 AM) rsinger: er, beginning
(11:29:09 AM) mjgiarlo: rsinger: you are begin not to sense make.
(11:29:12 AM) rsinger: edsu: what we hope to achieve, i mean
(11:29:20 AM) charper: edsu: thanks again for pulling all of this together.
(11:29:38 AM) charper: Great group of people to help push these concepts further into reality.
(11:29:45 AM) brocadedarkness: yes thanks a lot (even though i haven't really been paying much attention today... will do better next time)
(11:32:12 AM) rsinger: so, do we want a 'practical approaches to making your data more linked/linkable'?
(11:32:13 AM) lbjay: good lord, was that really 1.5 hours?
(11:32:50 AM) mmmmmrob: lbjay: nope, not coming this year
(11:33:29 AM) jphipps: edsu: thanks!
(11:33:35 AM) lbjay: mmmmmrob: too bad. i was thinking your word re: MARC -> RDF would be enlightening to some
(11:33:41 AM) lbjay: s/word/work/
(11:33:51 AM) rsinger: i.e. "so your data store is solr, you could approach it like this..."
(11:34:04 AM) mmmmmrob: lbjay: thanks - rsinger, iand and rjw all know about it
(11:34:21 AM) charper: rsinger: low-hanging fruit is good.
(11:34:34 AM) charper: Good example of what Martin's stuff does, too.
(11:34:41 AM) rsinger: yeah
(11:34:47 AM) charper: "You're data is in Aleph/Voyager - you could approach it..."
(11:34:55 AM) rsinger: exactly!
(11:34:59 AM) rsinger: dspace
(11:35:04 AM) mjgiarlo: whoever removed me from oai-ore on the wiki: thank you. :)
(11:35:11 AM) rsinger: omeka
(11:35:13 AM) charper: edsu: I think maybe registry/RDA should be split out into 2 topics?
(11:35:15 AM) rsinger: ,etc.
(11:35:25 AM) charper: rsinger: yep.
(11:35:26 AM) edsu: charper: can you split on the wiki?
(11:35:42 AM) ***mjgiarlo will be content to sit there and learn, really.
(11:35:43 AM) edsu: mjgiarlo: heheh, didn't want to put words in your mouth :)
(11:36:05 AM) rsinger: seems like registry and openvocab might be related?
(11:36:11 AM) charper: edsu: wiki's been split.
(11:36:52 AM) charper: Reason I suggest is that I'd like to discuss the path from low-hanging fruit to sky-pie...
(11:36:55 AM) dbs: miker_ will be there, he should have some ideas about LOD and Evergreen
(11:37:02 AM) edsu: charper++
(11:37:03 AM) mjgiarlo: charper: Is good.
(11:37:23 AM) rsinger: how about "create a foaf file and see why it's useful"?
(11:37:49 AM) charper: rsinger: foaf file for yourself, I'm assuming?
(11:38:16 AM) rsinger: charper: well, the agenda item is "create a foaf file"
(11:38:31 AM) mjgiarlo: foaf seems to be the canonical example. it's on the tbl page ref'ed earlier.
(11:38:33 AM) charper: rsinger: I ask because I think about Martin's use of foaf for giving info on the libraries in Sweden...
(11:38:51 AM) mjgiarlo: and qualifies as low-hanging
(11:38:55 AM) rsinger: charper: well, i would like to see the extension led there, yes
(11:39:10 AM) rsinger: mjgiarlo: right, but just like i don't need a pet store application...
(11:40:01 AM) charper: maybe that's the need for a sandbox.
(11:40:22 AM) charper: If we've got people creating foaf files and biblio descrips, do we want a bucket to toss them into?
(11:40:30 AM) charper: Provides something to play with during breakout sessions?
(11:42:41 AM) edsu: charper: great to see you in here btw :)
(11:42:58 AM) charper: edsu: thanks - I used to lurk here once in a while... I should more often. :)
(11:46:03 AM) ***mjgiarlo is still thinking of pet store applications
(11:46:22 AM) ***charper needs more coffee before noon meeting...
(11:54:22 AM) rsinger: heh, i think my point is being mistaken
(11:54:28 AM) rsinger: re: pet store
(11:54:50 AM) rsinger: my point is that it's generally not a simple leap for me from 'pet store' to 'application i actually want or need'
(11:58:13 AM) edsu: rsinger: i forgot to ask if you had something you were willing to talk about in the demo period for lod-preconf
(11:58:27 AM) rsinger: edsu: i am not sure, yet
(11:58:51 AM) rsinger: edsu: and i'm not sure it wouldn't just be better served with martin/anders
(11:59:25 AM) mjgiarlo: demos can be time-consuming anyhow.
(12:00:31 PM) mjgiarlo: I'd rather see three solid demos than seven rushed (or long) ones, in the interest of not inundating or confusing people.
(12:01:19 PM) lbjay: how many cooks does it take to bake a semweb sky-pie?
(12:01:19 PM) edsu: rsinger: i wonder if jangle fits in the discussion?
(12:01:44 PM) rsinger: not if we want to limit it to rdf or rdf/a
(12:02:10 PM) edsu: i'm kind of +0 about that limitation
(12:02:25 PM) ***rsinger nods.
(12:02:31 PM) edsu: i guess i think it would be great for "us" to have that discussion
(12:02:45 PM) rsinger: edsu: yeah, agree
(12:02:51 PM) edsu: but the danger is the attendees will scratch their chin and mutter wtf, and wander off to the oclc demo
(12:03:18 PM) rsinger: edsu: i think limiting scope on a clearly defined objective is a good thing
(12:03:27 PM) mjgiarlo: I think I'd be bored by debate.
(12:03:51 PM) edsu: mjgiarlo: yea
(12:04:12 PM) mjgiarlo: Tell me why I should care about linked data, and how to use it with my data, and I will be overjoyed. Assuming that is true of many, but, uh, yeah, "assuming."
(12:04:14 PM) rsinger: edsu: i mean, if jangle fits into the lod talk, so does unapi
(12:04:38 PM) edsu: jangle is a bit more web-like than unapi imho
(12:04:49 PM) edsu: from what i've seen anyhow
(12:04:57 PM) rsinger: mjgiarlo: that's exactly what i expect the crowd to be like
(12:05:19 PM) edsu: yeah, i guess they signed up because of what it said on the tutorial page
(12:05:24 PM) edsu: and that did say rdf and shit
(12:05:27 PM) rsinger: edsu: well, jangle is exactly what you were referring to when you mentioned 'what dchud wants sounds more like REST'
(12:05:44 PM) edsu: rsinger: aye
(12:05:51 PM) rsinger: since what dchud said sounded exactly like jangle
(12:06:04 PM) rsinger: plus there's a 20 minute talk on jangle the next day
(12:06:11 PM) rsinger: and i can propose a breakout session
(12:06:20 PM) edsu: rsinger: yes, i think this makes sense
(12:06:29 PM) rsinger: jangle's not hurting for exposure on that front
(12:06:32 PM) mjgiarlo: RDF is fine and all but I don't think we need to advocate RDF specifically, rather we should advocate linked data (which necessarily includes RDF in some form or another?).
(12:07:05 PM) mjgiarlo: I didn't word that well, but hopefully it made sense.
(12:07:12 PM) rsinger: i think this preconf should focus on using rdf
(12:07:30 PM) edsu: did you mean 'does not necessarily include' ?
(12:07:53 PM) edsu: the thing about rdf is that it has linkedness baked into it at a very low-level
(12:07:55 PM) mjgiarlo: rsinger: I guess what I mean is that we don't need to argue about the value of RDF, per se, but about the value of linked data.
(12:08:09 PM) rsinger: i don't think we need to argue about the value
(12:08:09 PM) edsu: i think things like html and atom etc have this linked-ness too tho
(12:08:10 PM) mjgiarlo: and carry RDF along with it where it makes sense.
(12:08:38 PM) rsinger: let's just assume that people saw that this was about linked data and rdf, like it says on the wiki, and take it from there
(12:08:50 PM) mjgiarlo: edsu: can you do linked data without any RDFisms? (genuine question, not a challenge.)
(12:09:19 PM) mjgiarlo: Yes yes.
(12:09:19 PM) edsu: mjgiarlo: depends on what you mean by linked-data i suppose
(12:09:27 PM) rsinger: well, i think you're just remaking rdf, but yeah
(12:09:45 PM) mjgiarlo: we ought to be clear and consistent on what we mean by linked data then :)
(12:10:36 PM) edsu: mjgiarlo: yeah, i mean it all comes down to the 4 guiding rules
(12:10:43 PM) edsu: mjgiarlo: for me
(12:10:43 PM) rsinger: i guess even the attempts to do so without rdf would require something like grddl to make it useful
(12:11:03 PM) mjgiarlo: 4_rules++
(12:11:12 PM) edsu: yes, grddl is pretty nice though
(12:11:19 PM) edsu: esp for xslt-heads
(12:11:42 PM) ***edsu shrugs
(12:12:08 PM) edsu: it may be kind of confusing, but being able to point at deployed linked data apps, and say this is what they did will help dispell confusion i think
(12:12:20 PM) edsu: that's when being about to point at the lod-cloud is important
(12:12:31 PM) edsu: it helps people understand i think
(12:13:01 PM) rsinger: edsu: i agree, but it's still turning whatever you've got (microformats, for example) into rdf
(12:13:02 PM) mjgiarlo: I see that point. My main concern is that we don't try to cram in too many demos.
(12:13:07 PM) rsinger: edsu: which means the end-result is still rdf
(12:13:39 PM) rsinger: although, i guess jangle meets timbl's 4 rules
(12:13:56 PM) edsu: rsinger: yes
(12:14:06 PM) rsinger: helps when the first two rules are really the same rule
(12:14:21 PM) edsu: mjgiarlo: even if people just talked for 10 mins?
(12:14:34 PM) edsu: mjgiarlo: to get people interested in a breakout session in the afternoon?
(12:15:14 PM) rsinger: i think it's important that the demos give context to the point at hand
(12:15:53 PM) rsinger: for instance, showing lcsh.info, and the underlying architecture doesn't prove the thesis terribly well
(12:16:32 PM) mjgiarlo: It's doable, as long as the pre-demo stuff is solid and provides aha-ification.
(12:16:49 PM) mjgiarlo: ... and as long as demo-ers don't get carried away. :)
(12:17:15 PM) edsu: yeah, we need to apply a realistic timetable to what we have
(12:17:24 PM) edsu: with breaks for coffee, etc
(12:17:25 PM) mjgiarlo: the first hour of the pre-conf is the most crucial.
(12:17:55 PM) edsu: i wonder if perhaps a general overview of linked data would be useful before diving into rdf
(12:17:59 PM) rsinger: yes
(12:18:03 PM) rsinger: edsu: definitely
(12:18:20 PM) rsinger: edsu: linked data, a couple of examples, THEN how to do it
(12:18:33 PM) edsu: i see someone added it already :)
(12:18:38 PM) mjgiarlo: I believe iand may have offered to do that?
(12:19:06 PM) edsu: yes, but he's going to do the busy_developer's_intro to rdf as well
(12:19:08 PM) mjgiarlo: I just added a couple bullets from cerealtom's doc is all
(12:19:18 PM) edsu: cool
(12:19:27 PM) mjgiarlo: I thought iand said he was going to massage his rdf piece into an intro to linked data?
(12:19:32 PM) mjgiarlo: I dunno, so hard to keep up.
(12:19:48 PM) edsu: the vocab discovery is in two places
(12:19:51 PM) edsu: lbjay's idea
(12:20:00 PM) edsu: mjgiarlo: he said he could yeah
(12:20:31 PM) edsu: i wonder if stefano would like to participate
(12:20:33 PM) lbjay: edsu: either works for me
(12:21:59 PM) mjgiarlo: lbjay: it'll depend on whether you see it more as a demo or as a topic, it seems
(12:24:14 PM) lbjay: mjgiarlo: yeah, just pondering that
(12:24:53 PM) edsu: the line between intro and demo is kind of blurry
(12:25:01 PM) lbjay: i see it as a "let's all work through these practical examples together" thing
(12:25:11 PM) edsu: lbjay: yea
(12:25:16 PM) edsu: lbjay: right after foaf i think
(12:25:35 PM) edsu: lbjay: or in place of foaf
(12:27:38 PM) lbjay: edsu: i'm not sure i understand the foaf part
(12:28:04 PM) edsu: it would amount to writing a foaf file for yourself
(12:28:22 PM) rsinger: right, i think there needs to be one more step
(12:28:22 PM) edsu: in turtle, or rdf/xml or something, or maybe even using one of the generators
(12:28:31 PM) rsinger: showing something cool with these new foafs
(12:28:46 PM) rsinger: otherwise it's like doing a worksheet in middle school
(12:29:31 PM) lbjay: would you be mine, could you be mine, won't you be my FOAF?
(12:29:33 PM) edsu: hmm yea
(12:29:58 PM) rsinger: "great, i wrote a pet store application"
(12:30:10 PM) rsinger: "now what the fuck do i do with all this marc data?"
(12:30:18 PM) edsu: i guess pointing at livejournal and identica wouldn't be enough?
(12:31:13 PM) rsinger: edsu: but what's the practical use of a foaf doc?
(12:33:00 PM) lbjay: rsinger: some kind of web of trust system?
(12:33:10 PM) edsu: rsinger: it's more an exercise yeah -- practically someone could discover a distributed social network, with http
(12:33:16 PM) rsinger: edsu: i think part of the challenge (and the payoff) is showing the benefits of making our data available in these ways
(12:33:53 PM) rsinger: and i think foaf is exactly the sort of thing that royt says, "well, great... now what do i do with this thing?"
(12:33:59 PM) rsinger: or dchud
(12:34:23 PM) rsinger: and i think that's a legitimate question
(12:34:52 PM) edsu: i guess being able to discover and maintain a distributed view of a social network isn't very useful?
(12:34:53 PM) rsinger: but, let's say we had a link to royt's foaf from his LJ column
(12:35:13 PM) rsinger: which then led to royt's other professional and creative output
(12:35:26 PM) mjgiarlo: edsu: if there was a way to -show- people how setting up a FOAF gets them that, I think rsinger and royt would be happy.
(12:35:38 PM) royt: pretty much
(12:36:45 PM) edsu: mjgiarlo: well someone setting up a foaf wouldn't see it, they'd only see it if they attempted to either manually or automatically follow the links
(12:36:46 PM) rsinger: edsu: does that make sense?
(12:36:55 PM) edsu: it does yeah
(12:37:02 PM) mjgiarlo: sort, uh, foaf network searcher + network visualizer sort of doodad.
(12:37:08 PM) edsu: why bother going through the effort of publishing data this way
(12:37:35 PM) rsinger: i think even just the simplest 'follow your nose' aspect
(12:37:40 PM) rsinger: staff directory
(12:37:56 PM) rsinger: or, in this case, attendee directory
(12:38:27 PM) rsinger: so the graph would show the intersections of foaf:made and foaf:project and all that
(12:38:58 PM) mjgiarlo: follow_your_nose++
(12:39:03 PM) edsu: mjgiarlo: so this isn't enough i take it: http://foaf-visualizer.org/?uri=http://lackoftalent.org/michael/foaf.rdf&hash=mjg
(12:39:30 PM) rsinger: right, but what's the context of foaf-visualizer?
(12:39:37 PM) mjgiarlo: edsu: if I could start there and follow my nose, maybe that'd be enough
(12:40:17 PM) rsinger: that's why i thought it might be interesting outcome to make a linked data app that represents the preconf
(12:41:48 PM) edsu: yeah, that's an interesting idea rsinger
(12:41:51 PM) rsinger: so the foafs become part of that
(12:42:08 PM) mjgiarlo: edsu: ah, neat, I like how your foaf links out to dbpedia
(12:43:14 PM) edsu: could use freebase now too i guess
(12:43:47 PM) edsu: i guess we could just tell everyone to create an identica account
(12:43:59 PM) edsu: and not bother with manually creating a foaf file
(12:44:13 PM) rsinger: well, we could leave it to their discretion
(12:44:48 PM) rsinger: personally, i'd rather my foaf be associated with dilettantes.code4lib.org than identi.ca
(12:44:51 PM) mjgiarlo: and they could always start with a template, like ldodds's foaf maker?
(12:45:43 PM) edsu: rsinger: you can have your cake and eat it too with owl:sameAs
(12:46:55 PM) edsu: anyhow
(12:47:15 PM) edsu: mjgiarlo: i think you are right about having 3 solid intro talks
(12:47:25 PM) edsu: and then opening things up to a bit more free form stuff
(12:48:27 PM) lbjay: well, if "solid" is a criteria that pushes me back to the free form part of the schedule
(12:48:58 PM) edsu: nah
(12:49:17 PM) lbjay: identi.ca needs a profile setting for an external foaf file
(12:49:51 PM) lbjay: edsu: i wasn't complaining
(12:50:27 PM) mjgiarlo: I think there'll be a good niche for vocabulary selection once the principles and value of LOD have been established. maybe folks will have seen (or started to tinker on their own) simple FOAFs and will be wondering "well, great, but I want to do something more sophisticated than look at mjgiarlo's ugly mug on foaf-visualizer. where do I get other predicates?"
(12:51:29 PM) lbjay: mjgiarlo: i agree. i think it's one of the more "fun" aspects of it too. but that might just be me
(12:51:57 PM) edsu: it is the fun part, kind of more along the lines of vocamp in spirit
(12:52:17 PM) edsu: i think the other library nerds will find it to be the fun part too
(12:53:15 PM) rsinger: yeah, that's why i was suggesting making the preconf into lod app
(12:53:23 PM) mjgiarlo: right, they may start to think "well, hey, maybe all this descriptive metadata stuff could be modeled in this way" (which would later be reinforced by jphipps' spiffy registry).
(12:53:36 PM) rsinger: so we have the individual talks, linked to who did them

Return to "LinkedData" page.