Difference between revisions of "Chatlog 2010-04-22 Data Catalog Vocabulary"
From W3C eGovernment Wiki
|Line 36:||Line 36:|
15:16:47 <sandro> Cory Casanave
15:16:47 <sandro> Cory Casanave
15:16:56 <vassilios> Vassilios Peristeras from the DERI team
15:16:56 <vassilios> Vassilios Peristeras from the DERI team
15:17:03 <sandro> Brand
15:17:03 <sandro> Brand
15:17:28 <sandro> Niklas Lindström
15:17:28 <sandro> Niklas Lindström
15:17:50 <sandro> Dan Thomas, DC
15:17:50 <sandro> Dan Thomas, DC
Revision as of 22:31, 25 April 2010
Please justify/explain all edits to this page, in your "edit summary" text.
<sandro> Guest: Kate Geyer <sandro> Guest: Erik (dret) Wilde <sandro> Guest: Jon Phipps <sandro> Guest: Peter Krantz <sandro> Guest: Luigi Montanez <sandro> Guest: Rich Wolverton <sandro> Guest: Niklas (lindstream) Lindström <sandro> Guest: William (ww) Waites <sandro> Guest: Dan (dan) Thomas <sandro> Guest: Dan (danbri) Brickley <sandro> chair: cygri <sandro> scribe: sandro <sandro> Present: Cory_Casanave, David_James, Ed_Summers, Fadi_Maali, George_Thomas, Martín_Álvarez, Richard_Cyganiak, Sandro_Hawke, Vassilios_Peristeras, Li_Ding, Brand_Niemann 15:10:46 <li> hi this is Li Ding from RPI 15:11:02 <li> I cannot join the telecon as I have a conflict telecon right now 15:11:24 <li> but I will stay in IRC, would there be anyone help scribing the conversation? 15:11:44 <sandro> topic: Introductions (several new people) 15:11:46 <Zakim> -Cory 15:12:09 <Zakim> +??P10 15:12:10 <sandro> sandro 15:12:12 <sandro> david james 15:12:39 <Cory> Zakim, ??p10 is Cory 15:12:39 <Zakim> +Cory; got it 15:12:40 <sandro> luigi 15:12:42 <sandro> Martín Álvarez 15:13:29 <sandro> George Thomas 15:13:59 <RichW> Rich Wolverton from Comm of Mass Supporting ODI 15:14:01 <sandro> Rich_Wolverton 15:14:52 <sandro> Erik Wilde 15:15:27 <sandro> Richard Cyganiak, 15:15:44 <sandro> Fadi Maali 15:16:01 <sandro> Kate Geyer 15:16:24 <sandro> Jon Phipps 15:16:47 <sandro> Cory Casanave 15:16:56 <vassilios> Vassilios Peristeras from the DERI team 15:17:03 <sandro> Brand Niemann 15:17:28 <sandro> Niklas Lindström 15:17:50 <sandro> Dan Thomas, DC 15:19:13 <sandro> William (ww) Waites 15:19:54 <sandro> next scribe: Ed Summers 15:20:16 <sandro> agenda: http://www.w3.org/egov/wiki/Data_Catalog_Vocabulary/2010-04-22 15:20:33 <sandro> topic: Presentations of existing work 15:21:21 <sandro> cygri: please keep it short -- 3 to 5 minutes, pointers to more details 15:21:27 <edsu> http://www.w3.org/egov/wiki/Data_Catalog_Vocabulary/DC-SKOS 15:21:37 <sandro> subtopic: Ed Summers, Brief Intro to Dublin Core and SKOS 15:22:11 <sandro> (page looks great, Ed) 15:22:25 <sandro> edsu: I imagine most folks are familiar with DC and SKOS 15:22:44 <sandro> edsu: DC started in 1995 to create a core set of elements to describe documents in the internet 15:23:06 <sandro> edsu: ... workshop in Dublin, Ohio. 13 elements. Meeting every years since then. 15:23:16 <sandro> ... turned into IETF and NISO/ISO standards 15:23:32 <sandro> ... co-evolved with RDF, due to people like Eric Miller 15:23:53 <sandro> ... DC has its own "Abstract Model", the way they make descriptions, very similar to RDF model 15:24:08 <sandro> ... expresses vocabulary in RDF, (DC Terms) 15:24:27 <sandro> ... I have links to how widely it's used. 15:24:42 <DavidJames> I'm having trouble logging into the wiki. When the meeting is over, perhaps someone can help? 15:25:39 <Zakim> -martin 15:25:41 <sandro> edsu: SKOS. 1999, mostly through EU projects. in 2004 moved to W3C. Jon Phipps was part of that. 15:25:58 <sandro> ... defined in "SKOS Reference" A W3C Recommendation in 2009. 15:26:42 <danbri> (the work started in DESIRE EU project - http://www.desire.org/html/research/deliverables/D3.6/d36b.html ... then Limber and SWAD-Europe. Thanks, EU taxpayers...) 15:26:44 <sandro> ... People with various controlled vocabs; skos lets them easily mint URIs for them, without re-engineering that whole information space (eg to fit into OWL). Lightweight way to get existing data of that sort onto the web. 15:27:02 <sandro> ... see examples on the web page. http://www.w3.org/egov/wiki/Data_Catalog_Vocabulary/DC-SKOS 15:27:40 <fadi_> http://www.w3.org/egov/wiki/Data_Catalog_Vocabulary/DERI_dcat_Demonstrator 15:27:43 <sandro> subtopic: DERI's dcat, Fadi 15:28:09 <sandro> fadi: dcat is an RDF vocabulary to represent government data catalogs. 15:28:26 <sandro> ... design was informed by analyzing existing catalogs. 15:28:40 <sandro> ... we looked at what attributes existing catalogs used 15:28:54 <sandro> ... then we looked at the data, to see how consistently it was used 15:29:06 <sandro> ... main entities are Dataset and Catalog 15:29:15 <ww> question re: dcat:granularity could perhaps be dcat:spatialGranularity and dcat:temporalGranularity? 15:29:39 <sandro> ... we tried to use skos:Concept 15:29:59 <sandro> zakim, who is talking? 15:30:11 <Zakim> sandro, listening for 10 seconds I heard sound from the following: ??P27 (13%), cygri (37%) 15:31:11 <lindstream> Re. distribution/updates: quite orthogonal. I like how dady extends void with that. 15:30:27 <sandro> subtopic: CTIC's Dataset Catalog Vocabulary and “meta-catalog” (Martin) 15:30:33 <sandro> martin??? 15:30:40 <sandro> zakim, where is martin? 15:30:40 <Zakim> sorry, sandro, I do not understand your question 15:30:42 <martin> I cannot connect to the conference, Zakim sais that it's full 15:30:54 <sandro> martin, press *0 to get the operator to add you 15:31:26 <sandro> if you can't get in, we'll do your presesntation next week. so sorry for not having enough space today. 15:31:43 <martin> I'm trying to contact an operator 15:31:41 <sandro> subtopic: Sunlight Labs API proposal (David, Luigi) 15:31:41 <LuigiMontanez> http://nationaldatacatalog.com/ 15:32:02 <sandro> DavidJames: National Data Catalog 15:32:03 <Zakim> +martin 15:32:15 <LuigiMontanez> http://wiki.sunlightlabs.com/Government_Data_Catalog_Guidelines 15:32:25 <sandro> DavidJames: We're looking for a few things. 15:32:30 <sandro> ... being ReSTful 15:32:50 <sandro> ... We view RDF as one way to express our data, but we're also looking for other formats 15:33:23 <sandro> ... developers really gravidate to lightweight solutions, and don't feel like RDF is worthwhile, from what they can tell. 15:33:43 <sandro> ... We're planning to add RDF support down the right, but at this point our key job is building buy-in 15:33:57 <sandro> ... We think the RDF community really understand how to make the data interoperable 15:34:12 <sandro> ... We want to adopt principals of RDF, even if we're not using the RDF formats right now. 15:34:41 <sandro> topic: Atom and dataset updates 15:34:47 <sandro> zakim, who is talking? #15:34:54 <Cory> Could whoever is typing mute please! 15:34:58 <Zakim> sandro, listening for 10 seconds I heard sound from the following: ??P27 (26%), cygri (60%), jonphipps (20%) 15:35:13 <sandro> cygri: What are the reasons we should consider other solutions? Are there things not easily covered by RDF? 15:35:23 <sandro> zakim, who is talking? 15:35:30 <edsu> uhoh 15:35:32 <LuigiMontanez> cannot hear 15:35:34 <Zakim> sandro, listening for 10 seconds I heard sound from the following: ??P27 (18%), cygri (49%) 15:35:36 <vassilios> +1 15:35:41 <dret> terrible noise 15:35:42 <sandro> zakim, mute ??P27 15:35:42 <Zakim> ??P27 should now be muted 15:35:48 <DavidJames> We are being jammed by touch tones 15:35:54 <edsu> damn! 15:36:05 <sandro> zakim, who is talking? 15:36:15 <Zakim> sandro, listening for 10 seconds I heard sound from the following: martin (3%), Sandro (46%), cygri (35%) 15:36:27 <sandro> q+ 15:37:48 <sandro> LuigiMontanez: It's mostly the tooling, as a developer. As a ruby or python developer. We're used to working with twitter or facebook API. 15:37:52 <dan> JSON is a notation scheme, RDF is structure. here is example of RDF in JSON notation: http://n2.talis.com/wiki/RDF_JSON_Specification 15:38:03 <sandro> ... they don't tread into RDF world, even though it might be good for them, too. 15:38:23 <sandro> ... it's not that RDF is more complicated; it's about the tooling. 15:38:37 <sandro> ... ckan is the only one I know of. 15:38:45 <sandro> ack sandro 15:38:49 <DavidJames> I just added a section called "Context / Philosophy" to http://wiki.sunlightlabs.com/Government_Data_Catalog_Guidelines 15:38:50 <ww> dan: we've experimented with some success storing rdf/json in mongodb 15:38:52 <sandro> http://code.google.com/p/linked-data-api/ 15:40:05 <lindstream> ww, cool 15:40:35 <sandro> sandro: linked-data-api might be a good solution here [details] 15:40:51 <sandro> cygri: are you thinking about govt agencies, or developer/users? 15:41:01 <sandro> Luigi: Both. 15:41:12 <dan> ww: similarly working on crime summary vocab using this approach: http://www.ipublic.org/wiki/Crime_summary_vocabulary 15:41:15 <sandro> Luigi: The govt agency usually just wants to make a website. 15:41:27 <sandro> ... they say "Well, how do we provide the data/api"? #15:41:32 <DavidJames> Actually, LuigiMontanez is talking :) 15:41:38 <sandro> ... we say use existing/popular techniques 15:41:44 <sandro> ... socrata [?] 15:42:00 <sandro> ... municipalities can get started very quickly that way 15:42:30 <sandro> ... at sunlight we try to convcince gocts to put their data on line. we want to tell them to do the easiest thing for them, and unforunately right now, that's not RDF. 15:42:59 <sandro> cygri: Quite a few catalogs are using simple structured forms, pre web, CSV. eg data.gov. 15:43:14 <sandro> cygri: So you want to go one step beyond this, eg json, xml. 15:43:37 <sandro> DavidJames: We're also looking for interop. I think that's where RDF really shines. 15:43:56 <sandro> DavidJames: let's create this RDF spec, but make it easy for agencies and developers to adopt it. 15:44:01 <sandro> +1 sounds great! #15:44:14 <sandro> [oops, that was Luigi, not David.] 15:44:40 <li> we have been playing around with CSV based catalog at data.gov 15:44:51 <sandro> DavidJames: Gov't agencies are ... somewhat behind. Permanent URLs would be great. Feeds would be great. 15:44:55 <Cory> +q 15:45:05 <sandro> DavidJames: THEN we can push for RDF, etc. There's a LOT to do. 15:45:17 <li> at RPI. and there are some imperfect entries 15:45:21 <sandro> DavidJames: We want to be careful about what we ask for. Short term + Long Term. 15:45:31 <sandro> ack DavidJames 15:45:39 <sandro> cygri: Great concrete requirements 15:45:50 <sandro> cygri:What do you mean "feeds". In which sense? 15:46:01 <li> for short time goal, we should consider work out a minimal set of vocabulary agreed by all 15:46:07 <li> and promote it 15:46:12 <sandro> DavidJames: Something that makes it easy for automatic importers, so it's easy to see when a record has changed. 15:46:39 <sandro> cygri: To be able to sync, to keep up with changes, which of 3000 data sets have changed in last week 15:47:09 <sandro> LuigiMontanez: Yes. To see which data sets have updated, and so developer knows when to pull and update their copy of the data. 15:47:14 <cygri> ack Cory 15:47:44 <George> you meant 'normal xml' 15:47:56 <sandro> Cory: In making RDF, in the short term, the striped data format often gets overlooked. RDF *CAN* be written (striped) to look like normal XML, with an XML schema, and processed with normal XML tools. But it's ALSO normal RDF. 15:48:07 <LuigiMontanez> http://www.stripesframework.org/display/stripes/Home 15:48:10 <sandro> +1 cory using constrained RDF/XML is nice. 15:48:21 <lindstream> have you seen: http://code.google.com/p/oort/wiki/Grit ? (self-promotion ;] ) 15:48:25 <sandro> George, I'm sure he did. 15:48:52 <LuigiMontanez> Oops, think my link was wrong 15:49:00 <sandro> cygri: yes, regular serialization is one approach. GRDDL is another way. 15:49:02 <lindstream> +q 15:49:04 <LuigiMontanez> This look more correct as to what Cory was referring: http://www.w3.org/2001/10/stripes/ 15:49:19 <sandro> cygri: Also RDFa 15:49:34 <sandro> cygri: Allows one to parse RDF graph out of HTML page. 15:49:38 <George> +1 Cory and cygri - developers need some schoolin' 15:49:39 <li> regarding to the changes in data.gov, we even use twitter to publish the changes https://twitter.com/datagovwiki 15:49:55 <sandro> cygri: Ways to bridge RDF and what developers want. 15:50:05 <sandro> ack lindstream 15:50:13 <edsu> lindstream: you on the call? 15:50:18 <sandro> lindstream.... 15:50:30 <lindstream> i hear you, you don't hear me? 15:50:31 <sandro> zakim, who is muted? 15:50:31 <Zakim> I see ??P27, Rich_Wolverton, +0789798aaee muted 15:50:56 <sandro> zakim, unmute ??P27 15:50:56 <Zakim> ??P27 should no longer be muted 15:51:05 <edsu> sandro++ 15:51:35 <Cory> luigu - yes, http://www.w3.org/2001/10/stripes/ is it - often overlooked as a bridge. What this doesn't say is that it can also have an XSD 15:51:59 <sandro> lindstream: Regarding formats, that's a very important point. Not trick people into producing RDF, but make it transparent for them. They do namespaced XML and it just happens to be RDF. That's what I proposed with GRIT, a normalized RDF. 15:52:13 <sandro> (I call this "rigid RDF" myself.) 15:52:24 <ww> support for serialising striped xml in the usual tools? 15:52:27 <li> AT RPI we have already experimenting RDFa for metadata publishing, here is an example http://data-gov.tw.rpi.edu/wiki/Dataset_34, and the parsed RDFa can be found at http://data-gov.tw.rpi.edu/ws/lodcx.php?url=http%3A%2F%2Fdata-gov.tw.rpi.edu%2Fwiki%2FDataset_34&operation=test. Further more, the RDFa data can be used to enrich search result, e.g. http://dat 15:52:34 <li> http://data-gov.tw.rpi.edu/ws/lodcs.php?query=earthquake 15:52:57 <sandro> cygri: note on IRC the RPI examples 15:53:05 <sandro> cygri: Opinions on RDFa? 15:53:27 <sandro> q+ to comment on dcat scope 15:53:38 <sandro> cygri: Is RDF acceptable to folks? 15:53:47 <sandro> zakim, mute ??P27 15:53:47 <Zakim> ??P27 should now be muted 15:53:56 <lindstream> I like RDFa a lot, but's afaik it can only be viewed with an "RDF eye" 15:54:01 <sandro> zakim, ??P27 is lindstream 15:54:01 <Zakim> +lindstream; got it 15:54:07 <sandro> lindstream, I muted you because of noise 15:54:42 <lindstream> (sandro, my line is very poor right now) 15:54:44 <sandro> ww: data.gov.uk rdfa is made by drupal front-end, it's completelye disconnected from backend rdf store #15:55:03 <vassilios> who said this? #15:55:05 <sandro> someone_else: that's unfortunately 15:55:13 <DavidJames> Luigi and I are taking a look at some fairly recent open source code for Ruby and RDF here: http://github.com/bendiken/rdf #15:55:55 <sandro> s/someone/William 15:56:04 <sandro> q? 15:56:09 <li> At RPI, we add an extension to semantic wiki to export the backend RDF into RDFa 15:56:33 <li> there are a little bit issues on usage of vocabulary, because RDFa consumers are more picky 15:56:39 <edsu> q+ to suggest dcat+rdfa note 15:56:48 <li> e.g. yahoo and google are taking different RDFa vocabulary 15:56:48 <sandro> cyg: is there somewhere easy to start, here? 15:57:10 <sandro> cyg: the updates/feeds issue was interesting. I don';t know how to handle that in pure RDF environment. 15:57:20 <sandro> q? 15:57:42 <li> if we want to get major search engines to take the RDFa, we should probably convince them the value of RDFa vocabulary <sandro> topic: Scope of this Work 15:57:56 <sandro> cygri: I also wanted to talk about scope for this group. What do we want to deliver, what are the artificacts? 15:58:08 <sandro> cyg: 1. Nail down a vocabulary, in a vocab reference document 15:58:27 <sandro> ... 2. Implementor's guide, telling folks how to use it in practice. 15:58:41 <sandro> ... 3. perhaps a Demonstrator 15:59:11 <sandro> ... 4. Use Cases? Requirements? Test Cases? Dunno if those are appropirate here. 15:59:18 <edsu> +1 15:59:21 <lindstream> Perhaps dcat should provide a separate "dcat distribution types" vocab? Or leave notification details out, and recommend (link to) existing or upcoming methods (AtomOwl, dady..)? 15:59:30 <sandro> proposed: extend by 15 mins 15:59:30 <Cory> +1 15:59:31 <sandro> +1 15:59:33 <DavidJames> +1 15 minute extension 15:59:36 <RichW> +1 15:59:37 <lindstream> + 1 15 min 15:59:38 <ww> +1 15:59:39 <kate_geyer> +1 15:59:39 <martin> +1 15:59:42 <vassilios> +1 15:59:46 <sandro> resolved: extend by 15 mins 15:59:47 <dan> +1 16:00:21 <lindstream> (not, leave out dcat:distribution, but the details it refers to) 16:01:19 <dret> +1 16:01:41 <sandro> sandro: let's focus on building an RDF vocabulary, here, and deal with the RDF-world bridge elsewhere. 16:01:41 <cygri> sandro: propose to focus on the RDF vocab in this group, being aware that there's more work to do afterwards 16:02:08 <sandro> sandro: also multiple implementions is better than one demonstrator. 16:02:12 <Zakim> - +1.202.564.aaii 16:02:20 <DavidJames> I like the idea of keeping this group focused on RDF. 16:02:22 <dret> what about patterns of how to expose certain services around the vocabulary in plain web ways? 16:02:33 <sandro> edsu: Yes, vocab note, and implementors guide. But also guide to how to use this outside of just RDF. 16:02:35 <li> at RPI, we would also like to contribute some thought on data.gov vocabularies 16:02:40 <DavidJames> Of course, for our work, we're interested in applying the principles to other formats as well. 16:02:42 <li> based on our experience, 16:02:44 <sandro> q? 16:02:47 <sandro> ack sandro 16:02:47 <Zakim> sandro, you wanted to comment on dcat scope 16:02:49 <sandro> ack edsu 16:02:49 <Zakim> edsu, you wanted to suggest dcat+rdfa note 16:03:28 <sandro> vassilios: Maybe we're mixing two discussions that we don't need to mix. 1. the vocab; 2. the format for the vocab. 16:03:41 <sandro> vassilios: Maybe have vocab in a neutral format, eg UML. 16:03:44 <dret> separating a data model and services provided around that vocabulary would be a very good idea. 16:04:05 <lindstream> Re. updates w. Atom or alternatives: see summary by Leigh Dodds: http://www.ldodds.com/blog/2010/04/rdf-dataset-notifications/ 16:04:27 <sandro> cyg: I'd be concerned about leaving concrete rep open, then we wont really get interop.) 16:04:34 <Cory> Not fully open - you need a closed set of rpresentations 16:04:49 <vassilios> +1 to cory 16:04:49 <sandro> cyg: Also, if we document a number of different ways, then, again, we don't get interop. 16:04:52 <dret> do you want to propose a standard, or best practices? i think this is a very important distinction to make 16:05:00 <sandro> +1 cygri interop is top priority. 16:05:49 <li> I think FOAF is a good case to follow 16:05:55 <sandro> erik: It's also important to have services. how you communicate updates, too. 16:06:27 <sandro> erik: this big problems: not just modeling the data, but how you use the data, and build services on top of it. 16:06:32 <sandro> ack DavidJames 16:06:55 <sandro> DavidJames: Where are people leaning on separating vocabulary from format 16:07:27 <lindstream> .. my thoughts on this topic are mostly at: http://code.google.com/p/court/ 16:07:36 <dret> vocabularies and servoces are two separate thongs, and both are essential for building interactions. 16:07:46 <dret> things, actually ;-) 16:08:01 <sandro> cyg: Difference between UML and RDF is fairly minor. 16:08:37 <Cory> We generaly do a UML model first, and produce RDF - but can also produce other formats 16:08:39 <ww> however it should be encouraged that e.g. json type services represent their data in the same shape as the rdf (or abstract) model 16:08:58 <Zakim> -lindstream 16:09:26 <George> +1 sandro 16:09:34 <li> if the vocabulary is fairly annotations and flat, we don't need a very complex description other than dublin core 16:09:44 <sandro> sandro: I don't see an advantage to going more abstract that what dcat does, RDF, with UML view of it. 16:09:47 <jonphipps> think abstract model is good idea, but not without concrete representations 16:09:51 <Cory> +1 on model first 16:09:53 <dret> servives? 16:10:36 <li> therefore, RDF rather than UML is good enough, I don't think Microformat vocabulary used UML for definition 16:10:48 <sandro> cyg: Think of the UML diagram in dcat, and imagine it without prefixes. Should we do that? Is that an important deliverable? 16:10:50 <dret> +1 16:10:56 <sandro> -1 ehhh 16:10:58 <kate_geyer> kate_geyer agrees with edsu 16:10:59 <li> the general principle is to keep everything as simple as possible 16:11:01 <George> -1 16:11:08 <fadi_> -1 16:11:11 <jonphipps> 1 edsu 16:11:21 <li> requirement users to understand UML could be extra cost 16:11:25 <DavidJames> +1 16:11:30 <George> +1 sandro 16:11:47 <sandro> sandro: Maybe this is mostly editorial, for later. 16:11:57 <vassilios> do you believe that users understand better rdf than uml? 16:11:58 <DavidJames> I agree, deriving a JSON representation isn't a big problem (for this group at least) 16:12:00 <dret> i like the idea of an abstract model, and would like to add a set of services around interactions. 16:12:21 <edsu> dret, so rdfs isn't good enough for the model? 16:12:25 <DavidJames> And there are multiple, interchangable RDF serializations. I don't think we should worry about which is "official" 16:12:30 <sandro> cygri: so, should we do an RDF schema for classes/properties ---- the diagram with boxes and arrows. An abstract model, we can layer on? 16:12:31 <lindstream> lindstream has joined #egov 16:12:36 <DavidJames> +! 16:12:37 <sandro> +1 (if I understood it right.) 16:12:38 <DavidJames> +1 16:12:39 <cygri> +1 16:12:43 <Cory> Yes, your model should be representation independent - 16:12:52 <vassilios> +1 to cory 16:12:56 <martin> +1 16:12:58 <vassilios> this is the point 16:13:13 <vassilios> to make it computer independent model according to MDA 16:13:29 <edsu> +1 for what sandro said, if uml means a nice human readable html page 16:13:29 <sandro> sandro: I'm understanding that we'll design it in RDFS terms, but also be aware of what it looks liek in UML terms. 16:13:39 <Cory> The adavantage of UML for this is mostly that it is simpler to understand and can then map to multiple formats 16:13:46 <sandro> Yes, human readable. Not UML software readable. 16:13:50 <sandro> q? 16:13:52 <DavidJames> vassilios: what do you mean by MDA? 16:14:02 <vassilios> Model Driven Architecture 16:14:10 <Cory> UML is software readable! 16:14:38 <George> this isn't an mda crowd - let the omg do that 16:14:50 <George> or someone like cory :) 16:14:55 <sandro> edsu: RDF is a suitable data modeling language, and the right one for this crowd. 16:14:57 <LuigiMontanez> LuigiMontanez has joined #egov 16:15:14 <dret> there are other structured data formats advocated by the w3c, with much wider availability of tooling and developers in the wild. 16:15:17 <lindstream> +1 on sticking to RDF for the modeling 16:15:39 <DavidJames> I am happy to go with the group consensus on how we model 16:15:40 <sandro> edsu: The toolchains around data formats ... people want to work on documents sometimes .... let's make sure there are examples with RDFa and JSON etc. 16:16:15 <Cory> I don't see RDF/UML as conflicting, they are ways to express the same underlying concepts. 16:16:17 <edsu> atom and rss too :-) 16:16:39 <George> +1 cygri 16:16:56 <jonphipps> +1 cygri 16:16:59 <sandro> cygri: We want to do a model thats independant from the concrete syntax or format that we might do in the end. We want to use RDFS as the modeling language. This DOES NOT mean folks have to use RDF/XML. It's about using RDFS as the way to construct our vocabulary. 16:17:00 <edsu> +1 16:17:04 <sandro> +1 cygri 16:17:07 <li> My point is not rejecting UML. For simplicity concern, we should have something simple enough for regular web developers to take 16:17:36 <dret> sorry for repeating myself, but i think having services is essential. this could also be abstract or concrete, but it would be very helpful as "best practices" for somebody trying to do this. 16:17:41 <DavidJames> I agree with modeling in RDF without thinking about particular serialization details. For this group, why should we care about the particular RDF serialization? 16:17:47 <jonphipps> +1 cygri: "using RDFS as the way to construct our vocabulary." 16:17:47 <sandro> cory: A UML model is easy for people to understand. The equivalen RDF rep, the equiv JSON rep, ... not a problem. 16:18:22 <sandro> cygri: is that agreement, Cory? 16:18:41 <DavidJames> +1 RDF schema 16:18:41 <sandro> cory: UML is more often used, but RDFS is okay. 16:18:54 <sandro> (time check) 16:18:58 <edsu> dret, i agree, i think i heard cygri addressing this in the 'demonstrator' piece? 16:19:14 <Zakim> -George 16:19:39 <sandro> cygri: Erik, you stress the imporance of services. Would you agree that having an abstract model in the first place is useful for defining those services? Or do you need the services first? 16:20:15 <sandro> dret: It would be good to have the model first. It might be hard to follow, if it's all abstract. "You might want to expose this bit of JSON", etc, easier to understand. 16:20:33 <sandro> dret: look at data model as static structures, AND servces as the way people interact with the data in that model. 16:20:35 <DavidJames> We already extended 15 minutes, so I think we should be aware of the time... 16:21:17 <edsu> sandro, sounds almost like the w3c egov social media compone 16:21:19 <sandro> dret: recovery.gov initial folks were urged to publish as feeds, and didn't know how to do it. Having best practices and validation tools, so folks can test if their services are well behaved or not, that's goo dfor developers. 16:21:28 <edsu> s/compone/component/ 16:21:31 <sandro> cygri: out of time, thanks everyone! <sandro> topic: Adjourn 16:21:43 <dret> thanks! 16:21:46 <DavidJames> Thanks 16:21:46 <RichW> RichW has left #egov 16:21:49 <kate_geyer> thanks, all! 16:21:49 <Zakim> -Sandro 16:21:51 <kate_geyer> kate_geyer has left #egov 16:21:53 <Zakim> - +0789798aaee 16:21:54 <Zakim> -Kate 16:21:57 <lindstream> Erik, how open-ended a service? I'm thinking "an algorithmic resource for slices of time and/or category".. (delivering doc updates and/or changesets) 16:21:59 <Zakim> -Rich_Wolverton 16:22:00 <Zakim> -Erik 16:22:01 <martin> thanks, bye 16:22:01 <Zakim> -??P48 16:22:02 <Zakim> -edsu 16:22:02 <Zakim> -Cory 16:22:09 <lindstream> thanks all 16:22:09 <vassilios> thanks, bye 16:22:10 <Zakim> -martin 16:22:11 <Zakim> -cygri 16:22:11 <Zakim> -jonphipps 16:22:28 <edsu> cygri, nice job :-) 16:22:53 <jonphipps> hmmm. having trouble logging into the wiki with w3 login 16:23:09 <jonphipps> anything different about it? 16:23:09 <dret> services can be whatever you like them to be. as simple as a feed with very static semantics, or as complex as a query language exposed somewhere. often, the best idea is to have something in the middle "form-based queries", so to speak. 16:23:11 <edsu> jonphipps, i think it got changed up recently 16:23:19 <Zakim> - +1.410.336.aahh 16:23:40 <cygri> martin, sorry for skipping your topic in the end! would you be willing to show it next time? 16:23:54 <dret> bye. 16:23:56 <martin> ok, no problem 16:24:07 <jonphipps> I was able to get onto the swdwg wiki earlier in the week 16:24:10 <edsu> jonphipps, http://lists.w3.org/Archives/Public/public-egov-ig/2010Apr/0024.html 16:24:24 <cygri> note to self: things always take twice as long as i'd like them to :-) 16:24:45 <jonphipps> oh, _very_ recently :-) 16:25:02 <jonphipps> btw, nicely done cygri 16:26:01 <edsu> fwiw, i added my dcat in atom strawman to the wiki http://www.w3.org/egov/wiki/Data_Catalog_Vocabulary/AtomStrawman 16:27:28 <lindstream> edsu, nice 16:28:06 <edsu> lindstream, i imagine i may have goofed some things up, please feel free to edit 16:28:20 <Zakim> disconnecting the lone participant, Sunlight, in T&S_EGOV(LD TECH)10:00AM 16:28:24 <Zakim> T&S_EGOV(LD TECH)10:00AM has ended 16:28:26 <Zakim> Attendees were Sandro, +1.202.955.aaaa, Luigi, David, [CTIC], +1.202.690.aabb, martin, +1.617.521.aacc, +1.510.643.aadd, +0789798aaee, +03539149aaff, cygri, fadi_, +1.617.619.aagg, 16:28:28 <Zakim> ... George, Erik, +1.410.336.aahh, +1.202.564.aaii, Rich_Wolverton, Kate, Cory, jonphipps, edsu, lindstream 16:29:31 <martin> cygri, thanks, I can present when you want 16:29:57 <lindstream> edsu, I'll take a closer look #16:30:55 <sandro> ciao, lindstream ! 16:31:06 <lindstream> lindstream has left #egov 16:31:07 <edsu> seeya lindstream 16:34:12 <DavidJames> DavidJames has left #egov 17:00:03 <LuigiMontanez_> LuigiMontanez_ has joined #egov 18:16:35 <cygri> cygri has joined #egov 18:25:50 <LuigiMontanez> LuigiMontanez has joined #egov 18:32:35 <Zakim> Zakim has left #egov # SPECIAL MARKER FOR CHATSYNC. DO NOT EDIT THIS LINE OR BELOW. SRCLINESUSED=00000530