13:15:25 <RRSAgent> RRSAgent has joined #ledp
13:15:25 <RRSAgent> logging to http://www.w3.org/2011/12/06-ledp-irc
13:15:36 <dbooth> zakim, who is here?
13:15:37 <Zakim> sorry, dbooth, I don't know what conference this is
13:15:39 <Zakim> On IRC I see RRSAgent, Zakim, timbl, dbooth, Cornelia, ericP
13:15:46 <dbooth> zakim, what conferences?
13:15:46 <Zakim> I see SW_(LEDP)7:30AM, UW_Team()8:00AM active and no others scheduled to start in the next 15 minutes
13:15:55 <dbooth> zakim, this is ledp
13:15:55 <Zakim> ok, dbooth; that matches SW_(LEDP)7:30AM
13:19:30 <Julius> Julius has joined #ledp
13:20:07 <bheitman> bheitman has joined #ledp
13:20:11 <Ralph> Ralph has joined #ledp
13:20:14 <martynas> martynas has joined #ledp
13:20:27 <Ralph> zakim, MIT-G449 is MeetingRoom
13:20:27 <Zakim> +MeetingRoom; got it
13:23:37 <SteveBattle> SteveBattle has joined #ledp
13:28:02 <dbooth> Tim: We need patterns.
13:28:55 <dbooth> ... Some use relative URIs, some absolute.
13:29:56 <julius> julius has joined #ledp
13:29:57 <ericP> scribenick: dbooth 
13:30:08 <dbooth> ... What are the patterns we use?  And which patterns have we not been using because we don't have the tools?
13:30:51 <labratmatt> labratmatt has joined #ledp
13:30:59 <dbooth> Topic: Introductions
13:31:38 <dbooth> EricP: I do mostly Life-Sci/healthcare in W3C. Same needs for enterprise data.
13:32:18 <dbooth> SteveSpiker: IBM.  Looking at linked data for integration approach, specificaallly for soft dev.
13:33:20 <dbooth> MartinNelly:  IBM, CTO of Rational brand.  My group came to LD looking for solution to app programming model.  Wanted a suite of apps that would work better together than in the past.  Discovered in LD a different way of looking at the problem that is promising.
13:33:40 <ericP> Arnaud Lehors: (IBM)
13:33:48 <dbooth> ArnaudDehors: IBM, Softward Stds group.  Current role is standards lead for LD.
13:33:50 <Zakim> +Ralph
13:34:01 <ericP> ... standards lead from IBM
13:34:06 <dbooth> TedSlater: Merck, interested in integrating lots of data from lots of places.
13:34:45 <dbooth> __: Korea, ____
13:35:03 <dbooth> ... Came here for DB integration.
13:35:57 <ted_> ted_ has joined #ledp
13:36:21 <dbooth> CorneliaDaivs: EMC, corporate CTO office.  Data int is a constant challenge.  Systems are siloed.  Challenge is going to the product groups saying we want them to think about LD, and they don't understand what it means.  Need to think about how to achieve aLD vision wo overwhelming the developers.
13:36:25 <sspeiche> sspeiche has joined #LEDP
13:36:44 <dbooth> ___: Harvard, innovation lab.  Integrating data across libraries, federated models.
13:37:18 <sandro> sandro has joined #ledp
13:37:26 <Arnaud> Arnaud has joined #ledp
13:37:38 <dbooth> John___: IBM, sys mgmt and operations.  Installations at telcoms, etc.  Looking at operationalizing or linking the data across apps.  Our large customers have this odd notation that they should be able to enter the data once and not care where it is.
13:38:01 <dbooth> SandroHawke: W3C, e-Gov lead, and a bunch of working groups, RIF, OWL, etc.
13:38:29 <dbooth> AshokMuhultra: Oracle, drumming the RDB-RDF for a while.
13:38:54 <dbooth> BenjaminHyghtman: Galway, research perspeective. 
13:39:25 <dbooth> DavidWood: 3 Round Stones, current project is CallimachusProject.org  creating a LD mgmt system to visualize LD across the web..
13:40:20 <sandro> DavidBooth: Interested in the pipeline for data generation, not homogeneous, number of steps.
13:40:36 <sandro> (skipping TimBL)
13:40:46 <ted_> LEDP on Lanyrd at http://lanyrd.com/2011/ledp/
13:40:52 <dbooth> Julius: New to LD, one projj w martinez. 
13:41:46 <Arnaud1> Arnaud1 has joined #ledp
13:41:49 <dbooth> Martinez: from Denmark, working w an instrument company building websites.  Also side projects using RESTful APIs.  Now blended those things into one and it went so well we made a new project for it.  Similar to Callimachus.
13:42:29 <dbooth> BradAllen: Elsivier. perspective of large ent w info solos.  Also as a publisher want to understand how to scale for customers across content sources across the web. 
13:43:16 <sandro> ericP, access on these minutes?    agreed to be public?
13:43:18 <dbooth> AlanDegoda: Elsivier, leading a pladform dev for LD to open access and discover to access content and leverage ext data sets.
13:44:02 <dbooth> SteveBattle: Workgin w a small startup in UK.   Also open source dev on Callimachus.  I'll be talking about taking components from Callimachus w Jena.
13:44:17 <bheitman> what kind of command should I use in order to provide my name and my affiliation for the scribe ? 
13:44:53 <dbooth> RalphSwick: W3C.  I'm happy to have you here and see progress.
13:45:14 <Zakim> -Ralph
13:45:30 <dbooth> rrsagent, make logs public
13:47:08 <dbooth> EricP: If i want to make some data avail, i can put it in pages, use RDFa, etc.  Or I can put my data in a sparql database.
13:48:48 <dbooth> TimBL: some are using RDf all the time, but only thinking about tables that they get from their sparql queries.  Not worried about the gory RDF.
13:49:11 <dbooth> Cornelia: Security is a big issue also.
13:50:27 <dbooth> Alan: We're increasingly getting into apps that use triple stores to do integration.  We have both use cases.  Evolutionariily we've been working from static toward dynamic from RD.
13:50:35 <Zakim> +Ralph
13:51:24 <dbooth> Martin: We've given up on tables.
13:51:25 <timbl> Zakim, who is on the call?
13:51:25 <Zakim> On the phone I see MeetingRoom, Ralph
13:52:15 <dbooth> ___: Pondered the last few years about how LD could be applied in big data and battling residual resistance to these technologies.  My position paper is on identity.
13:53:48 <Zakim> -Ralph
13:54:22 <sandro> BREAK UNTIL 9:15
13:55:01 <AndyS> AndyS has joined #ledp
13:56:12 <Zakim> +Ralph
13:56:30 <Alan> Alan has joined #ledp
13:56:51 <AlanYagoda> AlanYagoda has joined #ledp
13:56:52 <Zakim> -Ralph
14:00:21 <tlr> tlr has joined #ledp
14:01:50 <Zakim> +Lalana
14:02:06 <Ralph> zakim, Lalana is really Ralph
14:02:08 <Zakim> +Ralph; got it
14:02:27 <Ralph> zakim, Lalana is temporarily Ralph
14:02:27 <Zakim> +Ralph; got it
14:17:21 <ora> ora has joined #ledp
14:18:09 <dbooth> s/___:/OraLassila:/
14:18:30 <ora> Presentation: Linked Data as an Application Integration Architecture -- Martin Nally, IBM/Rational
14:18:35 <dbooth> Topic: First Presentation
14:19:12 <sandro> ashok: Shall we write a white paper?
14:19:13 <dbooth> Ashok: After hearing your stories, should we write a white paper saying what is the state of LD today?  What are people doing? Where is it working well?  Would you help write it?
14:19:33 <dbooth> (A few indicate yes)
14:19:46 <dbooth> Martin: Good start, but will need much more than that.
14:20:08 <ted_> I'll contribute.
14:21:10 <dbooth> Topic: Martin Nelly
14:22:48 <dbooth> Martin: IBM Rational.  Life looked good, but trying to evolve toward the web.  Saw pressure for more global dev, and customers were tired of separated tools.  Their processes go from one end to the other, and they want their tools working together that way.  This is our story, but there's nothing very special about this one.
14:23:15 <dbooth> ... Security is also important -- what's shared and what's not.  All the normal enterprise concerns.
14:24:03 <dbooth> .. People have been trying to integrate these tools for a long time, and the predominant most basic one is glue code within the tool to do point-to-point connections.  
14:24:54 <dbooth> ... It's worked for a long time, but it's endured because when you open an API others can script to it.  But it is tightly coupled.
14:25:08 <dbooth> ... What are the integration functions that people need?
14:26:06 <dbooth> ... Create a link betwen artifacts in different tools, create an artifact in another tool.  Share common concepts across tools, e.g., people, team, project, release.
14:26:23 <dbooth> ... Or be able to query across information in multiple tools.
14:28:18 <dbooth> ... People have been working on this problem for a long time.  Typical attempt is to try to use a single common repository.  Some of our competitors are still trying to do this, which makes me happy because I know how it will turn out -- the same as it's turned out all the other times we tried it.
14:28:34 <Arnaud> Arnaud has joined #ledp
14:29:38 <dbooth> ...  Or another is the ESB approach.  It's a bit more structure than the n^2 approach, but it still has the same basic disadvantages.  And the ESB becomes the bottleneck.  Hasn't worked out well for us.
14:30:04 <dbooth> ... We've  been stuck with these approaches for 20 years.
14:30:40 <dbooth> http://www.snopes.com/quotes/internet.asp
14:31:35 <dbooth> ... Linked data allows groups to work independently, loosely couples, tech neutral, minimalist, etc.
14:32:49 <ted__> ted__ has joined #ledp
14:32:50 <julius_> julius_ has joined #ledp
14:32:55 <dbooth> ... So in 2005 or 2006 we started adopting this style.  But on the cons: it was unproven when we started; big paradigm shift; lots of invention required.
14:33:31 <Arnaud1> Arnaud1 has joined #ledp
14:33:39 <dbooth> ... People do not know how to do this.  Most of the answers are out there, but knowing where it is and how to find them and how to put them together -- that's what's missing.
14:34:29 <dbooth> ... Big orgs, trying to get these people to move in this direction is a huge issue.
14:36:05 <dbooth> ... We did this to implment their own stuff, but most do not only use Rational.  So we started to share this w our partners.    OSLC and open community.  http://open-services.org/
14:36:35 <dbooth> ... Some of what's on that site is pretty good, and a few pieces make me cringe.
14:36:47 <dbooth> ... Due to lack of guidance or ignorance.
14:37:45 <dbooth> ... There are some mini-ontologies.  I hope that part endures, it has some value.  There's another part called core.open-services.org and I hope that part goes away.  That's an attempt to capture best practices.  Want to find the right home for it.
14:38:36 <dbooth> ... The day you stop using XML is like ending a bad relationship.  You realize how much it was screwing up your whole life.
14:38:52 <LeeF> LeeF has joined #ledp
14:39:29 <dbooth> ... I need help leading an org down the right path.
14:40:12 <dbooth> ... problem 1: creating data on the web.  This is a read-write paradigm.  We're creating linked data on the web, so we need creation and update protocol as well.
14:40:26 <dbooth> ... And how do I find the things that already exist?
14:40:45 <dbooth> ... And the core.open-services.org part is where we got it wrong.
14:41:16 <dbooth> ... If you start w RDF you need to start w a basic resource that you can POST to, and then be able to do a GET to find out what's been posted before.
14:41:39 <dbooth> ...  And tis is how we create collections -- things w the same subject and predicate.
14:42:50 <dbooth> (Showed RDF example)
14:43:12 <dbooth> Martin: POSTing to this creates a resource w the same subject.
14:46:21 <dbooth> ... If you start from RDF, and you can POST to and GET from them, then that's all  you need!  
14:46:52 <dbooth> ... Very simple.  But you need something like this, otherwise everyone invents their own thing.
14:47:26 <martynas> martynas has joined #ledp
14:47:40 <timbl> RWLD
14:48:04 <dbooth> ... I first POST testCase1
14:48:21 <SteveBattle> Callimachus containers give you a very similar construct.
14:48:57 <dbooth> LeeFeigenbaum: Have you looked at the SPARQL 1.1 Update Protocol?  It does this sort of thing.
14:49:09 <sandro> Lee: Have you looked at the SPARQL 1.1 Graph Store Protocol
14:49:14 <sandro> Martin: No, we don't do SPARQL
14:49:15 <martynas> what about using sioc:Container / sioc:has_container? we're using it for a similar purpose (I think)
14:49:35 <sandro> Lee: It has SPARQL in the title, but isn't SPARQL
14:49:39 <dbooth> TimBL: In the client space domain, you could also PUT that graph.  Have you looked at the R/W LD thing?
14:49:45 <sandro> Martin: Then why does it have SPARQL in the title?
14:50:06 <dbooth> ... If you POST as media type Turtle then it will be appended to the graph.
14:50:23 <LeeF> s/SPARQL 1.1 Update Protocol/SPARQL 1.1 Graph Store Protocol
14:51:18 <dbooth> Martin: I POSTed testCase1, and as a side effect a triple got added to this graph.
14:51:28 <sandro> LeeF, Andy and I (the reviewers) both said the current title is okay, FWIW.    (And I said we should take the word SPARQL out of the title.)
14:51:42 <sandro> er, NOT OKAY
14:52:06 <LeeF> sandro, ***sigh***
14:52:29 <sandro> (and this interchange with Martin is why.  I've had this conversation acouple times, myself.)
14:52:31 <dbooth> TimBL: How did someone know that this is the way to POST there?
14:53:02 <dbooth> Cornelia: How do you know what you can POST to, and what format you can POST to?
14:53:21 <sandro> ericP, can/do we have a way to postpone/list issues like this?
14:54:08 <dbooth> Martin: I have these special URLs, and if you POST to them then it adds a resource and adds the triples.  I could write the whole spec on a napkin.
14:54:49 <ericP> sandro, sure, but i feel like this gets people into a useful mode
14:55:14 <dbooth> ... I would like a base spec for this.
14:56:03 <dbooth> TimBL: We have various people coding up the RW web stuff and they just got to this.  If something ends with a slash, should it have this property, like a directory on a file space?
14:56:04 <Cornelia> Cornelia: AtomPub addresses what you can POST to and what it is you are POSTing.
14:56:37 <dbooth> Martin: There are lots of domain models.  File system is one.
14:57:02 <dbooth> TimBL: In a secure env w access control, when you POST, that implies things about the access.
14:57:08 <LeeF> Elias Torres will be glad to know that 6 years later IBM is still working on matching up RDF with Ato/APP :-)
14:57:24 <dbooth> Martin: We have access control, but not a universal design that could become a std.
14:59:38 <julius> 30mins, but we started earlier
15:00:14 <dbooth> Martin: It took us 5 years to get to the harder parts.  Most of it wasn't inventing anything, but finding the right things to do.
15:00:40 <Zakim> +??P19
15:00:45 <dbooth> ... So if you take this approach, then you end up w thousands or millions of triples, so you need to paginate.
15:01:15 <dbooth> ... So for every URL we have another URL w "?nextPage" added to it.
15:01:34 <dbooth> ... This is ok if you don't care about the order of the triples.
15:02:21 <dbooth> EricP: How does "?next", "?next" avoid serving the same triples as "?firstPage"?
15:03:11 <ericP> dbooth: does ?nextPage assume there's a session?
15:03:19 <ericP> Arnaud1: it's in the RDF
15:04:13 <dbooth> Martin: RDF contains :testCases?firstPage = ...
15:04:50 <dbooth> ... and at the end ti says: :testCases?page2 bp:nextPage rdf:nil .
15:05:18 <martynas> why not solve pagination just by mapping query parameters like ?ofsset=10&limit=20... directly to OFFSET, LIMIT etc. in SPARQL?
15:05:41 <dbooth> EricP: When does the ?page2 triple get into the RDF?
15:06:51 <sandro> ( Since this is for the client's limits, I think the client should pick the page size, ?pagesize=<n>  )
15:06:57 <dbooth> TimBL: If someone asks for the whole thing, do you see the need for an HTTP response saying "I've kind of given you what you want, but not the whole thing", or a 303?
15:07:24 <dbooth> Martin: Yes, it does a 30x.
15:08:21 <dbooth> Ora: I can't help feeling like this is conflating user interface issues?  Shouldn't the client say how much it can consume?
15:08:53 <sandro> martin:   syn is ?pageSize=<N>
15:08:55 <dbooth> Martin: There's also a synonym for ?firstPage where you can say how much data you want.
15:09:33 <sandro> SteveB: LinkedDataAPI does this differently...
15:09:47 <dbooth> TimBL: There will be people who say that you should use HTTP for slicing because it does that already.
15:11:32 <dbooth> Cornelia: We've dealt w this paging thing (using ATOM), but order always matters.
15:11:34 <Zakim> -Ralph
15:12:26 <dbooth> Martinez: Seems natural to map these things to things we have in SPARQL.
15:12:40 <dbooth> Martin: This is pre-SPARQL.  Very simple.
15:13:06 <dbooth> TimBL: But using the same terms will make it easier for people to grok.
15:14:04 <SteveBattle> The Linked Data API allows you to set number of resources per page as a configuration parameter. Clearly, pagination is a key issue with (HTML views of) linked-data.
15:14:09 <dbooth> Martin: we often have metadata about the resources, and we only want to GEt the metadata.
15:15:23 <John> John has joined #ledp
15:15:27 <dbooth> ... We have also proposed "Basic Profile for LD"
15:16:36 <dbooth> ... All of our resources use RDF.  
15:16:45 <sandro> Martin: The big problem with RDF/XML is that it gets in the way of trying to help people understand how RDF is different from XML.
15:17:03 <dbooth> ... RDF/XML is bad not only because it is so ugly, but because it makes it harder to get people away from XML.
15:17:53 <dbooth> (Martin outlines 12 rules for Basic Profile for LD)
15:20:19 <dbooth> Martin: Open World Assumption (OWA) is one of the hardest things for people to get used to.
15:20:59 <dbooth> ... Closed world is what leads to monstrosities like UML.
15:21:56 <dbooth> ... Programmers want to test everything.  But then nobody can create anything anymore.
15:22:35 <dbooth> ... People also get very elaborate in making their links.  Instead, have people represent links as predicates.
15:22:53 <dbooth> ... And if you need to add qualifiers or annotations they should be more triiples.
15:23:37 <dbooth> ... And sometimes they even reify RDF or make their own reified-like node.
15:24:27 <dbooth> Ora: I am very happy to see rule #9 and want it as a bumper sticker, but it's a hard sell.
15:25:14 <SteveSpeicher> Here's a reference to what Martin referenced: http://www.ibm.com/developerworks/rational/library/basic-profile-linked-data/index.html
15:25:27 <dbooth> dbooth: This list is a great start -- straw man for basic profile.
15:25:50 <SteveBattle> David: We need to take this spec into the alley and beat it up a little. :)
15:26:49 <ale_de_vries> ale_de_vries has joined #ledp
15:27:41 <dbooth> Martin: Need "If it wants to be a good duck, it must walk like one and quack like one"
15:27:57 <dbooth> TimBL: To avoid spam, might need a duck filter.  :)
15:28:24 <LeeF> For the record & for those who are not aware, Clark & Parsia have worked a great deal on semantics for integrity constraints with OWL 
15:29:21 <dbooth> Martin: Classic validation and constraints are tricky -- people don't know how to do it.
15:29:23 <martynas> SPIN uses SPARQL to define rules and constraints http://spinrdf.org/spin.html
15:29:47 <dbooth> ...  Don't infer that Mary is the same person as Jane, tell me there's an error.
15:30:19 <timbl> Theer are two models, one in whohc you truyst your apps and you must give then access to write arbitrary stuff into the storage, and the other in which you say have an unathenticated feed to pub;ic notifications and you wanrt them to be limite dto exclude spam and be filterered to allow a simple announcement triple.
15:30:30 <bheitman> just FYI, the current speaker is now starting to cut into the timing of the original schedule
15:30:44 <dbooth> ... PUT doesn't work well for updating data.
15:30:57 <dbooth> ... SQL has no equivalent of PUT.
15:31:21 <dbooth> ... Proposed solution: PATCH
15:33:09 <dbooth> TimBL: Have you looked at the design issues note on RW data?  In some cases you're using WebDAV server and you dont' have an option.  But if yo uhave a smart server, a SPARQL update resource is POSTed
15:33:48 <timbl> http://www.w3.org/ReadWriteLinkedData
15:34:39 <timbl> ooops
15:34:43 <martynas> this returns 404 :)
15:34:46 <julius> I think this is right one: http://www.w3.org/DesignIssues/ReadWriteLinkedData.html
15:34:58 <timbl> 1) http://www.w3.org/DesignIssues/LinkedData.html
15:35:12 <timbl> 2) http://www.w3.org/DesignIssues/ReadWriteLinkedData.html
15:35:16 <sandro> Also see: http://www.w3.org/2009/sparql/docs/http-rdf-update/#http-patch     WHICH YOU SHOULD COMMENT ON, IF YOU DONT WANT IT TO BE A W3C STANDARD SOON!
15:35:25 <timbl> 3) http://www.w3.org/DesignIssues/CloudStorage.html
15:35:26 <dbooth> Martin: Teaching people to have a more resource-centric view (instead of desktop-app-in-the-browser) is hard.
15:35:34 <timbl> (That's an ordreed list)
15:36:49 <dbooth> Martin's paper: http://www.w3.org/2011/09/LinkedData/ledp2011_submission_7.pdf
15:37:07 <dbooth> Topic: Graphity
15:37:26 <sandro> Zakim, who is on the phone?
15:37:26 <Zakim> On the phone I see MeetingRoom, ??P19
15:37:49 <dbooth> zakim, who is here?
15:37:49 <Zakim> On the phone I see MeetingRoom, ??P19
15:37:50 <Zakim> On IRC I see ale_de_vries, John, martynas, LeeF, Arnaud, julius, ted__, ora, tlr, AlanYagoda, AndyS, sandro, SteveSpeicher, labratmatt, SteveBattle, Ralph, bheitman, RRSAgent,
15:37:52 <Zakim> ... Zakim, timbl, dbooth, Cornelia, ericP
15:38:00 <dbooth> scribenick: SteveSpeicher
15:39:05 <dbooth> http://www.w3.org/2011/09/LinkedData/ledp2011_submission_1.pdf
15:39:15 <SteveSpeicher> Topic: Graphity
15:39:59 <julius> link to slides: http://www.slideshare.net/seporaitis/graphity-generic-linked-data-platform
15:40:00 <SteveSpeicher> Submitted position paper: http://www.w3.org/2011/09/LinkedData/ledp2011_submission_1.pdf
15:40:36 <SteveSpeicher> Paper title: "Graphity – A Generic Linked Data Platform"
15:40:57 <mcdonoug> mcdonoug has joined #ledp
15:42:50 <ballen> ballen has joined #ledp
15:43:15 <Arnaud> Martynas Jusevicius
15:43:33 <sandro> sandro has changed the topic to: Agenda: http://www.w3.org/2011/09/LinkedData/Schedule
15:43:37 <sandro> sandro has changed the topic to: Agenda: http://www.w3.org/2011/09/LinkedData/Schedule
15:43:43 <dbooth> s/Martinez/Martynas/g
15:43:53 <sandro> sandro has changed the topic to: Agenda: http://www.w3.org/2011/09/LinkedData/Schedule
15:45:25 <SteveSpeicher> martynas: Giving overview of comic Danish website and content on the site as http://heltnormalt.dk
15:46:02 <SteveSpeicher> …look to rebrand the website in 2011 for all the various supported content types
15:47:12 <SteveSpeicher> …overview of old code base, model contained many details of data represented on website
15:47:12 <dbooth> s/…/.../G
15:47:47 <SteveSpeicher> …model became very bloated over time
15:49:14 <SteveSpeicher> …reference of EJB model as leaky abstractions in comparison
15:49:19 <Sumalaika> Sumalaika has joined #ledp
15:50:08 <SteveSpeicher> …issues with many incompatible APIs, similar to what was referenced in previous session
15:51:16 <SteveSpeicher> …many different data (model) conversions, everything needed to connect to everything including model mismatch
15:51:56 <John> John has joined #ledp
15:52:39 <SteveSpeicher> …highlight the fact that if data sources were based on standard model (linked data), there would be 0 data conversions
15:54:06 <SteveSpeicher> …linked data picture gives a nice picture but doesn't explain how data is managed inside the bubbles
15:55:20 <SteveSpeicher> …REST+RDF: built off layers of abstractions and existing work in JAX-RS, as well as concepts from Jena…using PHP so can't use Jena
15:56:09 <SteveSpeicher> …defined an abstract RDFResource class
15:57:03 <SteveSpeicher> …in the end came up with a simple model for the site and platform, using data store to host the model (or any other supporting data)
15:57:12 <ericP> http://www.slideshare.net/seporaitis/graphity-generic-linked-data-platform slide 12
15:57:39 <Zakim> -??P19
15:58:08 <SteveSpeicher> …using XML works well for us, including processing RDF/XML and processing via XSLT
15:58:54 <sandro> !   (I'm surprised to hear RDF/XML + XSLT working.   Doesn't that break under arbitrary differences in the RDF/XML, because of a software update on the SPARQL software, or something...?)
15:59:02 <SteveSpeicher> …use RDF/POST encoding as well for the form post
16:00:01 <SteveSpeicher> …not only have RDF flowing in/out of their linked data "bubble" but also between all internal components
16:00:12 <ericP> http://www.slideshare.net/seporaitis/graphity-generic-linked-data-platform slide 13
16:01:16 <SteveSpeicher> …have data conversions = 2, using relational model mapping to RDF and DOM(?)
16:01:55 <John> sandro, it appears they are always building the DOM themselves; RDF is their lingua franca, so presumably they'd import into an RDF store, represent that as an RDF/XML DOM and then XSLT the DOM they built
16:02:07 <Sumalaika> Is there a formal W3C group working on the JSON-LD?
16:02:32 <SteveSpeicher> …look to reuse as much as possible, even if not directly PHP (like JAX-RS)
16:02:57 <SteveSpeicher> Sumalaika, there is a community group on it
16:03:15 <Arnaud> the RDF WG is officially in charge
16:03:34 <Arnaud> but the work is currently taking place in a community group
16:04:03 <Arnaud> at some point the RDF WG will look at the possibility of taking their spec and turning it into a Working Draft
16:04:36 <Sumalaika> Thank you Arnaud and SteveSpeicher
16:04:37 <SteveSpeicher> martynas: codebase comparison, summary new platform working with 15 content types and old one handled just 1
16:05:02 <ericP> i note that XSLT is way more verbose than PGP
16:05:07 <SteveSpeicher> …code trimmed down an order of magnitude, not many bugs as a result
16:05:09 <ericP> s/PGP/PHP
16:06:07 <SteveSpeicher> …server load after deploying new model (draws an ahhh sound from room) showing much lower load
16:07:02 <SteveSpeicher> …provided some caching details (which I did not capture)
16:08:27 <SteveSpeicher> …sources of RDF come from a number of places
16:09:33 <SteveSpeicher> …when blending generic linked data, UI and various RDF sources can start creating interesting mashups
16:09:48 <ballen> JSON-LD info at http://json-ld.org/
16:10:46 <SteveSpeicher> …suggested W3C involvement: RDF/JSON, RDF API, GRDDL stylesheet repo
16:11:01 <ballen> Mailing list at http://lists.w3.org/Archives/Public/public-linked-json/
16:11:48 <SteveSpeicher> …GRDDL looks nice but have problems finding stylesheets somewhere
16:12:06 <dbooth> Martynas: GRDDL is nice, but how do you find GRDDL XSLT stylesheets? 
16:12:35 <SteveSpeicher> …more community involvement by moving to something like http://w3c.github.com
16:13:42 <SteveBattle> Sorry to advertize - but Gloze is one of my open-source tools to map between XML and RDF without fiddly XSLT stylesheets <http://code.google.com/p/gloze/downloads/detail?name=Gloze-1.01-beta2.zip>
16:14:25 <presbrey> presbrey has joined #ledp
16:14:37 <bheitman> details for the next talk: 
16:14:38 <bheitman> Topic: Lessons and requirements learned from a decade of deployed Semantic Web applications
16:14:38 <bheitman> Speaker: Benjamin Heitmann, Digital Enterprise Research Institute (DERI), NUI Galway, Ireland
16:14:38 <bheitman> Co-authors: Richard Cyganiak, Conor Hayes, Stefan Decker (all DERI)
16:14:38 <bheitman> Slides: http://www.slideshare.net/metaman/w3c-ledpslides
16:14:39 <bheitman> Position paper: http://www.w3.org/2011/09/LinkedData/ledp2011_submission_2.pdf
16:14:40 <bheitman> Full journal paper: http://tinyurl.com/semweblessons (requires IEEE access)
16:14:58 <dbooth> Martin: What would be a ref arch for an RDF app?
16:15:14 <SteveSpeicher> Martin: in JEE brought a framework, model and thing be to build off of, would this time (Graphity) might be an interesting reference platform for LD/RDF applications
16:16:28 <SteveSpeicher> martynas: components depend on http/rdf in most cases and logic can exist in almost any programming language
16:16:40 <domel> domel has joined #ledp
16:16:42 <melvster> melvster has joined #ledp
16:17:32 <SteveSpeicher> bheitman: Topic: Lessons and requirements learned from a decade of deployed Semantic Web applications
16:18:43 <timbl> RRSAgent, pointer?
16:18:43 <RRSAgent> See http://www.w3.org/2011/12/06-ledp-irc#T16-18-43
16:18:56 <martynas_> martynas_ has joined #ledp
16:19:16 <SteveSpeicher> …talk is about some items around reference architecture and decade of semweb appplications
16:19:51 <martynas_> I think the RDF/POST encoding is undeservedly unknown: http://www.lsrn.org/semweb/rdfpost.html Allows you to encode RDF with standard HTML forms
16:20:03 <SteveSpeicher> …started with Scientific American article
16:21:00 <SteveSpeicher> …survey of over 100 applications over 10 years
16:23:19 <SteveSpeicher> …source from Semantic Web Challenges 2003-2009, ESWC 2006-2009, 12 questions asked, own analysis of paper
16:24:03 <SteveSpeicher> …analyzed paper by self/bheitman (not own application author)
16:24:22 <martynas> martynas has joined #ledp
16:25:05 <SteveSpeicher> …65% validated entries (responses to email validation), problem perhaps based on short life of academic email addresses
16:27:45 <SteveSpeicher> …highlights class of applications, standards and vocals used (summary in slides)
16:30:13 <SteveSpeicher> …conceptual arch has community consensus with 3 main components: RDF handling, data integration UI
16:31:35 <SteveSpeicher> …large number of apps only show data (read-only) vs creating/updating
16:32:39 <SteveSpeicher> ericP: what classes of apps are these?  for example for enterprise use
16:33:09 <SteveSpeicher> bheitman:: use a class of components in their applications used in enterprise apps (if I captured that right)
16:33:58 <martynas> FYI: uploaded my latest presentation as .pdf: http://semantic-web.dk/presentations/LEDP2011.pdf
16:34:18 <SteveSpeicher> …all apps consume LD but in incompatible ways
16:35:05 <SteveSpeicher> …embedding RDF improved with RDFa
16:35:30 <SteveSpeicher> Martin: can you explain more between more with import and export?
16:36:49 <SteveSpeicher> bheitman: summarized typical cases of having to manually export some files out of apps to then have to import the file into others
16:37:37 <SteveSpeicher> …LD gaps, writing RDF data is hard and 71% of apps don't support update/write
16:39:20 <SteveSpeicher> …distribution of app logic: many comps and standards, distributed hard to coordinate
16:40:10 <SteveSpeicher> …typically 3 data models in apps (graph, relational, oo): results in many roundtripping issues resulting in loss of data
16:40:47 <ballen> Slides for Elsevier talk: http://dl.dropbox.com/u/2641069/LEDP2011%20v3.pdf
16:41:03 <SteveSpeicher> …need more guidelines, best practices and design patterns…researchers don't receive this well due to sw engineer process don't apply
16:42:19 <SteveSpeicher> …need more sw libs beyond RDF storage, good libs can help reinforce guidelines/patterns
16:43:55 <SteveSpeicher> …another sw eng solution is sw factories: need the components and patterns to enable this
16:44:46 <SteveSpeicher> …full article at http://tinyurl.com/semweblessons
16:45:48 <SteveSpeicher> Out of apps surveyed, did you look at how many dereferenced URIs vs SPARQL?
16:45:50 <SteveSpeicher> bheitman: no
16:46:28 <davidwood> davidwood has joined #ledp
16:46:44 <SteveSpeicher> dbooth: Any observations that are not part of this survey which seemed focused on isolated cases, like enterprise?
16:47:07 <LeeF> s/Out of apps/LeeF: Out of apps/
16:47:46 <SteveSpeicher> bheitman: no clear class of these apps, can see some patterns that are typical patterns in enterprise apps in the apps here
16:49:12 <ericP> ericP has changed the topic to: http://dl.dropbox.com/u/2641069/LEDP2011%20v3.pdf
16:49:20 <Arnaud> Bradley Allen: http://www.w3.org/2011/09/LinkedData/ledp2011_submission_11.pdf
16:49:47 <SteveSpeicher> scribenick: Arnaud
16:49:50 <Arnaud> use of linked data for scientific publishing
16:50:53 <Arnaud> moving from a document centric approach to a linked data based approach
16:52:33 <Arnaud> trying to figure out how to go from acquisition of information in traditional documents as well as new social media and turning into "research objects"
16:52:35 <Zakim> +[IPcaller]
16:53:42 <Arnaud> lots of legacy processing that is XML centric
16:54:02 <Arnaud> trying to move that to linked data is a real challenge from a cultural and infrastructure point of view
16:54:38 <domel> domel has joined #ledp
16:55:05 <Arnaud> move also requires buy-in from suppliers of content
16:55:37 <Arnaud> ld design patterns for application development is key
16:56:29 <Arnaud> started from a standards point of view
16:56:47 <Arnaud> culture is based on validation of information
16:57:09 <Arnaud> use xml containers for rdf content against which validation is done
16:58:24 <Arnaud> consolidating existing xml taxonomies in skos
16:59:44 <Arnaud> developed an infrastructure around an ld repository with crud api and atom feeds
17:00:22 <ericP> ora clarifying slide 5 bullet 1
17:00:43 <Arnaud> enabling specialized applications with semantic search, such as for medical
17:02:46 <Arnaud> publishers are very interested in the kind of mashups that is now possible, like lancet
17:03:30 <erickn> erickn has joined #ledp
17:03:54 <Arnaud> need tools and best practices for managing URLs
17:05:13 <Arnaud> also need best practices for publishing and consuming ld
17:05:28 <Arnaud> various "definitions" of ld principles is confusing
17:06:04 <Arnaud> lack of guidance on how to serialize
17:06:49 <Arnaud> are we coding against RDF APIs or SPARQL?
17:06:58 <Arnaud> what about HTTP Range 4?
17:07:04 <Arnaud> Range 14
17:08:01 <Arnaud> also need toold and best practices for knowledge orgnization system management (although probably out of scope here)
17:08:28 <Arnaud> need validators for ld
17:08:28 <ericP> [slide 9]
17:08:57 <Arnaud> RDFa distiller is very useful, we need more of those tools
17:09:29 <Arnaud> need standards for: named graphs, version, access & entitlement
17:09:48 <Arnaud> annotation is also important, what standards should be used for that?
17:10:35 <Arnaud> there isn't much talk about free text search, but that's been critical for the web and needs to be addressed for ld
17:10:59 <Arnaud> martynas: do you use xslt?
17:11:13 <Arnaud> some
17:11:40 <sandro> brad: Named Graph == A document containing RDF that has a presence on the Web, as a container for that graph.
17:12:10 <ted__> The document's URI is used as the name of the named graph.
17:13:39 <Zakim> -[IPcaller]
17:30:34 <bheitman> bheitman has joined #ledp
17:45:18 <ballen> ballen has joined #ledp
17:46:48 <AndyS> AndyS has joined #ledp
17:48:30 <Zakim> +??P2
17:52:11 <bheitman> bheitman has joined #ledp
17:54:47 <dbooth> dbooth has joined #ledp
17:58:16 <John> John has joined #ledp
18:02:25 <ted_> ted_ has joined #ledp
18:04:32 <ted_> For the record, Ora's Save Ferris t-shirt is awesome.
18:05:35 <Sumalaika> Now we talk about managing a graph store and temporary graphs
18:05:47 <dbooth> Topic: Managing a graph store / Temporary RDF graphs
18:05:49 <dbooth> http://www.w3.org/2011/09/LinkedData/ledp2011_submission_6.pdf
18:05:56 <dbooth> http://www.w3.org/2011/09/LinkedData/ledp2011_submission_12.pdf
18:05:57 <Sumalaika> Dominik Tomaszuk the author is absent
18:06:06 <dbooth> scribenick: Sumalaika
18:06:31 <Sumalaika> The paper describes each resource as its own SPARQL end point  
18:07:15 <Sumalaika> David Wood presents the paper
18:07:36 <Nally> Nally has joined #ledp
18:07:52 <timbl> timbl has joined #ledp
18:08:01 <ted_> ted_ has joined #ledp
18:08:13 <Sumalaika> Uses HTTP methods to do CRUD on the resources
18:08:37 <Sumalaika> The slides/paper includes syntax for a temporary graph
18:09:45 <Sumalaika> Examples of triples are provided as URLs 
18:10:48 <Sumalaika> http://example.org/-person|rdf:tytpe|foaf:person/-person|foaf:name|-name/
18:11:20 <Sumalaika> constraining a triple (a la SPARQL) in a URL 
18:11:50 <Sumalaika> Generates a resource that is temporary .....
18:12:22 <dbooth> s/John___:/JohnArwe:/
18:13:53 <dbooth> s/___: Harvard/MatthewPhillips: Harvard/
18:14:04 <Sumalaika> Now John Arwe on Un-Cool URIs 
18:14:22 <dbooth> Topic: John Arwe on Un-Cool URIs
18:14:30 <domel> I am Dominik Tomaszuk. I wish that I could not come. David, thank you for showing my presentation.
18:14:31 <dbooth> http://www.w3.org/2011/09/LinkedData/ledp2011_submission_5.pdf
18:14:58 <Sumalaika> Your presentation looked very interesting Dominik - We miss you here
18:15:13 <davidwood> davidwood has joined #ledp
18:15:41 <tlr> tlr has joined #ledp
18:15:51 <davidwood> Domel, no problem.  I hope I didn't mangle your thoughts.  It would have been good to have you here.
18:15:57 <Sumalaika> Back to John Arwe ... he is discussing Ucncool URIs - URIs that change
18:17:57 <Sumalaika> Thre are many reasons for URIs changing, including business reasons such as project/server name changes, companies that get acquired, projects/servers get moved to other groups or locations ....   
18:18:41 <Sumalaika> John talks about:  
18:19:27 <Sumalaika> - Durable URL capable - so a DNS alias will suffice
18:20:09 <Sumalaika> - Redirect capable - so 301 server will suffice
18:21:08 <Sumalaika> URL prefix-mapping-capable - so provide URL prefix mappings to client and server components
18:22:15 <Sumalaika> A discussion ensues about guidelines for Linked Data ..... e.g., Linked Data hosting profile contract
18:23:35 <Sumalaika> Point is made about a social component -  maybe more implementations should support relative URLs (Best Practice for Software) 
18:24:52 <Sumalaika> So relative URIs goes up on the list on the  whiteboard for missing guidelines - specs (there are quite a few items on the list already) 
18:25:53 <Sumalaika> "LD profile" as well as "LD service" levels are on the summary list whiteboard
18:26:54 <Sumalaika> Now John discusses redirect awareness - and te performance impact when there aremultiple redirects
18:28:01 <dbooth> http://www.w3.org/TR/powder-dr/
18:28:19 <dbooth> http://www.w3.org/2007/powder/
18:28:35 <Sumalaika> We are told about the powder spec which allows the definition of URLs of a particular shape 
18:29:27 <Sumalaika> John tells about the "URL Oracle" concept that knows about all the real current URLs
18:33:13 <Sumalaika> John tells about servers that want to be moved (and don't mess up their own URLs) - a discussion ensuesas to whether relative URLs solve the problem ... the concept of URL groups ... the concept of internal infrastructure URLs that are not externally visible was also discussed 
18:34:15 <Sumalaika> There was a discussion about Oracle and Sun documemtation - and maintaining the Sun documentation under the Sun URL 
18:34:40 <sandro> sandro: sometimes I think it's best to use URLs like ns4343.com so they can be managed separately.
18:34:55 <Sumalaika> A discussion about PURLs : It is pointed out that LD products ship - it is the customers that have to manage the PURLs
18:34:55 <sandro> tim: Oracle should keep the sun.com URLs intact, where they are used
18:35:06 <sandro> ( I see http://java.sun.com/docs/books/jls/second_edition/html/jTOC.doc.html still working )
18:37:56 <julius> I'm making a list of keywords and list I find interesting, might find useful later: https://gist.github.com/1439307
18:38:59 <Sumalaika> A long discussion about the Linked Data discipline that Linked Data products impose on their customers  
18:39:05 <Sumalaika> on their users
18:40:25 <Sumalaika> Now John covers the scenario where someone re-uses an old URL or server name 
18:44:16 <Sumalaika> Jon talks about standards and their role with URL
18:44:47 <Sumalaika> A discussion ensues about the need for JSON RDF for UI components
18:45:29 <Sumalaika> SPARQL query results is a table - JSON  
18:46:44 <Sumalaika> A discussion about scraping JSON feeds - language simpler than GRDDL 
18:48:08 <Zakim> -??P2
18:48:25 <martynas> I've actually started making a generic XSLT transformation for RDF/XML > JSON-LD
18:49:00 <ted_> David Wood presents 3 Round Stones : diverted URI patterns.
18:49:22 <jma45> jma45 has joined #ledp
18:49:45 <ted_> ...motivated by mirroring of LD.
18:50:40 <ted_> scribenick: ted_ 
18:51:56 <ted_> ...routing of URIs to handlers, four different cases explained
18:55:12 <ted_> ..."diverted;" used as a token in these patterns
18:58:11 <ted_> ...multiple implications discussed, positive and negative
18:59:19 <ted_> ...Callimachus web patterns at callimachusproject.org
19:00:15 <ted_> ...another pattern in use is embedding SPARQL queries in Turtle, so that the queries can be named.
19:02:03 <ted_> ...useful for driving Google Chart widgets.
19:03:55 <ted_> ...Spin suggested, as was the Linked Data API from the UK.
19:06:19 <martynas> SPIN vocab can be used to construct SPARQL queries from RDF fragments
19:07:47 <ted_> ...demo of a mash-up of US nuclear power plant data
19:09:29 <ted_> ...demo.3roundstones.net
19:11:41 <ted_> ...Cambridge Semantics and Revelytix have come up with similar solutions.
19:13:30 <ted_> ...this pattern is about assigning a URI to the interface that is different from the one for the data
19:17:14 <ted_> ...Tim wants to empower the user to quickly choose from within the browser the particular view of the data they want
19:18:58 <ted_> ..."ensure the data doesn't die in the browser"
19:20:36 <Sumalaika> is uri opaqueness essential to Linked Data?
19:20:40 <martynas> try to look at this pattern as a Linked Data browser/proxy running as Web application in a normal Web browser: what you "type" in after /diverted is what you would type into the address bar; there is also caching involved, possibly history etc
19:22:33 <ted_> ... .wellkni
19:22:43 <ted_> Oops
19:23:03 <ted_> ... .wellknown is too limited to work here
19:25:24 <dbooth> Topic: Identity Crisis in Linked Data
19:25:36 <ted_> Ora Lassila et al 
19:25:38 <dbooth> by Ora Lassila (Nokia), Ryan Mcdonough (Nokia) and Susan Malaika (IBM)
19:25:46 <dbooth> http://www.w3.org/2011/09/LinkedData/ledp2011_submission_13.pdf
19:26:26 <dbooth> i/David Wood presents/Topic: Diverted URI Pattern
19:26:42 <ted_> ...background on web architecture and identity, E
19:27:51 <ted_> ...identity vs location, missing or ambiguous identity, versioning of data and identity, lack of stable identity
19:28:08 <ted_> ... Fire alarm!
19:28:50 <Sumalaika> Alert continues
19:29:01 <ted_> ... Identity vs Location
19:32:11 <dbooth> To clarify: there is a fire alarm in the building, but not in this *part* of the building.  This part of the building was not directed to evacuate at present.
19:32:21 <ted_> ...many different query URLs can yield the same resource
19:33:53 <ted_> ...Ryan is presenting this bit
19:35:46 <ted_> ... Internal identifiers and "RESTish" queries don't help, nor do URL queries with URI or URN identifiers
19:38:09 <ted_> ... Problem 2: missing or ambiguous identity (Ora again)
19:40:43 <ted_> ... Problem 3: versioning of data and identity, largely ignored by the W3C
19:41:10 <ted_> ...should version info be part of the identity of an object?
19:41:47 <ted_> ... problem 4:  lack of stable identity
19:42:00 <ballen> ballen has joined #ledp
19:42:30 <ted_> ..."cool URIs do not change" except they do
19:43:51 <ted_> ... 
19:45:01 <ted_> ...conclusions: confusion in matters of identity hinders interoperability. No particular solutions herr
19:45:30 <ted_> ... Er, here. That last was for everyone, regardless of gender
19:50:32 <ted_> ...lots of discussion RE versioning
19:51:30 <ted_> ...Ralph Hodgson has apparently created a versioning ontology (according to ericP)
19:52:15 <ted_> ..."identity crisis" should be a breakout topic
20:02:47 <ted_> Hilarious, dbooth!
20:03:46 <ted_> ... Tim came up with the Chief Identity Ofic
20:03:54 <ted_> Officer role
20:07:52 <sandro> ?: Maybe LD clients need to have access to search engines, for fixing broken links
20:08:27 <sandro> timbl: seach engine is the wrong model.   I think you don't want false positives; use crypto/signing.
20:08:36 <sandro> timbl: You could have enterprise catalogs.
20:08:50 <sandro> timbl: Fallback in a well-defined way.
20:09:30 <sandro> dbooth: Put a unique string in a document, so you can find it.
20:12:14 <sandro> timbl: Important to be able to 'follow your nose', start with JUST a URI and find everything you need.   If you do it that way, then you're always getting authoritative stuff, and immune to span.
20:12:20 <sandro> BREAK FOR 20 MINUTES
20:12:24 <dbooth> The folks who worked on ARK did a good job of pointing out he inherent problems in including *any* human-relevant info in an identifier: http://en.wikipedia.org/wiki/Archival_Resource_Key
20:13:10 <ted__> ted__ has joined #ledp
20:30:26 <ora> ora has joined #ledp
20:39:15 <Ralph> Ralph has joined #ledp
20:40:38 <dbooth> zakim, who is on the phone?
20:40:38 <Zakim> On the phone I see MeetingRoom
20:41:19 <dbooth> Topic: 	Achieving Linked Enterprise Data with RESTful Services and Link Relations
20:41:35 <dbooth> by Cornelia Davis (EMC)
20:41:49 <mcdonoug> http://www.w3.org/2011/09/LinkedData/ledp2011_submission_4.pdf
20:42:18 <mcdonoug> Cornelia Davis is on...
20:42:39 <dbooth> scribenick: mcdonoug
20:43:45 <mcdonoug> …being pragmatic about semantic web technology, rdf, owl, etc.
20:44:25 <mcdonoug> talking about examples of data services at EMC (slide 3)
20:46:29 <mcdonoug> …trendiness of the term "REST" and educating customers
20:47:02 <mcdonoug> …describing the principles of REST (slide 4)
20:48:40 <mcdonoug> …highlighting hypermedia as one of the biggest hurdles in REST
20:49:01 <mcdonoug> …talking about the state of existing frameworks (slide 5)
20:49:38 <mcdonoug> …no framework assists developers with hypermedia
20:50:00 <mcdonoug> …talking about ATOM (slide 6)
20:50:31 <mcdonoug> …(slide 7) ATOM already has support for links
20:51:00 <mcdonoug> …a links rel value indicates the semantics of a particular link
20:52:02 <mcdonoug> …(slide 8) detailing an example of ATOM
20:52:54 <mcdonoug> relating ATOM elememts to RDF
20:53:52 <mcdonoug> (slide 10) suggesting leverage patterns that relate to the semantic web but avoid using RDF
20:54:49 <mcdonoug> open for questions
20:56:06 <sandro> tim: People ask why RDF/XML, well, this is there to make this stuff not so strange  for folks coming from some angle.
20:56:27 <sandro> tim: the damage is, you make something look like XML, and then it doesn't actually make sense as XML.
20:56:54 <sandro> tim: people use Atom for RDF, then you can't actually load it into a triplestore.  :-(    I tried this with oData.
20:58:10 <sandro> tim: There were a lot of people for whom JSON was easy because they knew Javascript;....   stuff looks weird from different perspectives.
20:59:01 <sandro> martin: I'm very sympathetic.  We went through all of this.   We couldn't sell RDF, so we spent 2 years in hybrid land, trying to please both sides.    If you write your XML this way, both sides can kind of handle it.
20:59:13 <sandro> ... But it did not work out.    
20:59:25 <mcdonoug> ditto, hybrid solutions don't work
20:59:31 <sandro> ... People would continue to invent things that ended up making no sense.   RDF already did those things,.
21:00:12 <sandro> martin: So, 2 years of transition, 1 year of cleanup, but I'm not sure there was a shortcut.  Maybe that's the just the price to pay for switching a lot of people over.
21:00:28 <sandro> martin: Atom is a decent spec if XML is your starting point.
21:00:48 <mcdonoug> Xml gurus try to parse RDF/XML as XML
21:00:48 <dbooth> FWIW, I have seen XML gurus try to parse and understand RDF/XML as XML and it was a disaster.
21:00:54 <sandro> dbooth: I've seen XML gurus trying to consume RDF/XML and it was a distaster, because of the wrong mindset
21:01:41 <sandro> brad: A lot of what we've tried to do has been the same -- go to things people are comfortable with, and make the bridge.   That being said, I'm not sure I agree.   
21:02:30 <sandro> brad: What's important is RDF as a model, not RDF as a serialization.  If there's a way to leverage infrastructure, ...
21:02:56 <sandro> brad: How do we leverage ATOM, etc, to do things RDF doesnt do, like Pagination??   We've struggled witht hat.
21:03:12 <sandro> cornelia: Where did the transformation from oData to RDF fail?
21:04:29 <sandro> timbl: There were peices of Halo that were cleanly properties of a table, and I knew how to map those.   Then there were some links between web pages which stand for pieces of the table, in a way which I could not figure out how to map those, with a consistent semantics, repeatable, in the payload data.  There was not a clean boundary between the relational payload and the links in the outer peice.
21:04:57 <sandro> timbl: maybe I just hadn't got it, but after spending a day or so, I figured there wasn't a clean model.  But maybe someone else can do it.
21:05:47 <sandro> timbl: Clearly they were exposing an RDBMS, and I know how to do that in RDF, so as long as their mapping was reversiible, it's doable, but I couldn't figure out their mapping.
21:06:15 <mcdonoug> if we can't success on the polical problem, we can't succeed on the technical problems
21:06:29 <sandro> martin: I'm most concerned about Cornelia's problem.    How can i convince the 400K people at IBM?
21:06:29 <mcdonoug> RDF has been around for over  adecade, no explosive growth yet
21:07:03 <mcdonoug> david: fought the fight as a consultant form the outside in
21:07:37 <mcdonoug> the people who have been most succeful with don't attemp to the mapping to RDF at all
21:07:57 <sandro> davidw: I've fought this fight as a consultant.  As an author, I get around to fight it, too.  The people who have been most successful dont try to ease the path or do the mapping, as you just showed.  The problem is that there are lots of ways to solve any given problem.   You can take any given example application and the DB guy will do it one way, and the Web Services guy will do it, and the RDF guy will do it.
21:08:00 <mcdonoug> everyone wants to solve the same problem with a different hammer
21:08:23 <sandro> davidw: Where RDF really shines is in crossing silos, connecting things where traditional approaches have left off.
21:09:20 <sandro> davidw: Some orgs that have succeeded well (DoD, O'Reilly), they built a new team and hire ontologists if they need them, they get consultants in, they build a skunk works to do that bit between the silos.  They leave the DBAs in place, because the DBA stuff still needs to get done.
21:09:55 <mcdonoug> don't take silo people making them into data integration guys
21:10:05 <sandro> davidw: And they have consultants/new team to build out that bridging infrastructure.  You're not going to convert your silo folks -- really good at silos -- into data integration folks.
21:10:42 <sandro> Allen: That's what we're doing, with a startup group, showing we can solve this interop problem.
21:11:18 <sandro> Allen: When people see this, they perk up, and want to know more.
21:11:27 <cmatheus> cmatheus has joined #ledp
21:11:34 <mcdonoug> Silo developers are the ones implementing the atom interface on the top
21:12:06 <sandro> martin: Stick with the very simple stuff.   RDF triples and REST resources.
21:14:32 <sandro> martin: Even though this is very simple, the consequences are not.   Linked Data 101 -- complete enough to write real software, but keep complex stuff out.
21:14:38 <Arnaud1> Arnaud1 has joined #ledp
21:16:31 <LeeF> LeeF has joined #ledp
21:16:41 <mcdonoug> mcdonoug has joined #ledp
21:16:42 <Cornelia> Cornelia has joined #ledp
21:17:07 <Zakim> +??P0
21:17:09 <Zakim> -??P0
21:17:18 <mcdonoug> scibe has spotty internet connection...
21:17:19 <LeeF_> LeeF_ has joined #ledp
21:17:26 <julius> julius has joined #ledp
21:17:27 <Zakim> +??P0
21:17:35 <mcdonoug> http://www.w3.org/2011/09/LinkedData/ledp2011_submission_9.pdf
21:18:36 <sandro> <sandro> ora: I continue to be amazed how focused people are on syntax.        We were told lots of XML tools would help with RDF.     False.     I'm concerned the same problem with happen with JSON.
21:18:39 <sandro> <sandro> davidw: I completely agree.
21:18:40 <sandro> <sandro> eric cuts off conversation, as over time.   save it for a breakout (formats, outreach, sweet spot for learning.)
21:18:58 <sandro> Topic: Steve Battle, Linked Data API
21:20:03 <mcdonoug> mcdonoug has joined #ledp
21:20:55 <LeeF> LeeF has joined #ledp
21:21:02 <mcdonoug> should really be called "Linked Data API Framework"
21:21:35 <mcdonoug> uses Callimachus for presentation
21:22:28 <mcdonoug> showing HTML with RDF in Attributes
21:22:55 <mcdonoug> leverages link relations as well
21:24:07 <mcdonoug> alternatives to RDFa, Microdata, Microformats
21:24:29 <mcdonoug> Question: "what is the motivator to use RDFa?"
21:26:05 <mcdonoug> 1st motivation is SEO. 
21:26:22 <mcdonoug> search engines that read RDFa data can provide enchance search results
21:26:56 <mcdonoug> Google is the biggest consumer of RDFa
21:28:07 <Cornelia> Cornelia has joined #ledp
21:28:12 <martynas> martynas has joined #ledp
21:28:57 <mcdonoug> BestBuy, Facebook using RDFa in some form
21:30:26 <mcdonoug> RDFa as a web-template language...
21:31:57 <Sumalaika> Sumalaika has joined #ledp
21:32:54 <mcdonoug> legal HTML with RDFa elements.
21:34:40 <mcdonoug> arguing the if the template is legal RDFa syntax
21:35:06 <sandro> or, really, agreement it's the RDFa syntax, but observations that the semantics are different, because of variables.
21:35:34 <mcdonoug> tim: if you serve this up, it should have different mime type
21:36:00 <mcdonoug> describing the processing model of the Linked Data API...
21:36:09 <sspeiche> sspeiche has joined #LEDP
21:36:59 <mcdonoug> selectors will do pagination
21:37:35 <mcdonoug> viewer will process the template and compile it to SPARQL...
21:39:32 <mcdonoug> uses SPARQL OPTIONAL in queries in teh event that a given property is not present
21:40:23 <mcdonoug> allows automation of RDFs, avoids the use of XSLT, doesn't require the developer to learn SPARQL
21:40:34 <mcdonoug> LDA brings pagination to RDFa templating
21:41:03 <mcdonoug> …other groups working on RDFa presentation engines
21:41:52 <mcdonoug> can you use CONSTRUCT queries?
21:43:06 <mcdonoug> tim: do folks have strong optinions about XQuery
21:43:15 <mcdonoug> sorry: opinions
21:43:22 <dbooth> http://www.w3.org/Submission/2009/01/
21:44:42 <mcdonoug> Matthew Phillips: Federated Linked Library Data
21:44:50 <mcdonoug> http://www.w3.org/2011/09/LinkedData/ledp2011_submission_14.pdf
21:45:23 <dbooth> i/Matthew Phillips: Federated Linked Library Data/Topic: Matthew Phillips: Federated Linked Library Data
21:48:21 <mcdonoug> describing use cases for libraries and how linked data might address them
21:48:52 <mcdonoug> LibraryCloud - 15 million bibliographic records
21:49:42 <mcdonoug> tow thoughts: aggregate and host everything or federate the data across systems
21:50:23 <mcdonoug> is federated SPARQL feasible
21:51:28 <mcdonoug> q: what's in a bibliographic record?
21:53:32 <mcdonoug> licensing issues
21:53:48 <Zakim> -??P0
21:53:53 <SteveBattle> Slides for the "Extending the Linked Data API with RDFa" talk at <http://www.stevebattle.me/resources/LEDP2011.pdf>
21:54:03 <mcdonoug> while it's feasible with SPARQL federation, it might have issues
21:54:50 <mcdonoug> World Cats has this data, suggested that if you could get access to this data it may solve the problem
21:54:57 <martynas> FluidOperations have an advanced SPARQL federation tool called FedX: http://www.fluidops.com/FedX/
21:54:58 <mcdonoug> or some of the problems anyway
21:58:34 <mcdonoug> library community is slow on the update of new technologies
21:58:43 <mcdonoug> any solution would need to be super simple
21:59:54 <mcdonoug> Potential solical problems with libraries already using OCLC
22:01:02 <mcdonoug> organizing tomorrow's breakout sessions
22:03:32 <labratmatt> labratmatt has joined #ledp
22:03:54 <julius> RRSAgent pointer?
22:04:01 <julius> :(
22:08:59 <sandro> POSSIBLE-BREAKOUT: ROI WhitePaper
22:13:40 <julius> Compiled list of todays slides & various mentioned names/keywords: https://gist.github.com/1439307
22:30:18 <sspeiche> sspeiche has joined #LEDP
22:35:01 <Zakim> disconnecting the lone participant, MeetingRoom, in SW_(LEDP)7:30AM
22:35:05 <Zakim> SW_(LEDP)7:30AM has ended
22:35:07 <Zakim> Attendees were EricP, MeetingRoom, Ralph, [IPcaller]
23:01:16 <Zakim> Zakim has left #ledp
23:05:35 <mcdonoug> mcdonoug has joined #ledp
23:13:03 <Sumalaika> Sumalaika has joined #ledp
23:46:01 <LeeF> LeeF has joined #ledp