See also: IRC log
<ivan> last minutes
RESOLVED accept minutes http://www.w3.org/2012/01/11-swcg-minutes.html
Next meeting: 22 February, David to scribe
RDFa: short last call. Hope to get to last call in 3-4 weeks.
Provenance WG: new draft for data model.
ivan: RDD-to-RDF Working Group: close to candidate recommendation.
ivan: Two HTML data TF drafts: Jeni on Microdata and
RDFa and Gregg on RDF mapping. No major comments, so probably will be published as
notes.
... Two F2F meetings since last call. GLD and Provenance.
<pgroth> http://www.w3.org/blog/SW/2012/02/05/report-provenance-working-group-2nd-face-to-face-meeting/
Paul: as in blog post.... we had good F2F.
... Have gotten alot of public WDs out, but work around consensus on model.
... So timeline has slipped 2-3 months.
...Issue: first data model was quite complicated. Feedback on need to simplify.
But still too complicated.
...Paul: encouraged to simplify. Trying. But complications to simplification
process. E.g., agreement on 1-2 constructs.
... At meeting, why we were having complications. Addressing two use cases:
... 1) provenance statements about things on the Web - simple statements
... 2) more complex provenance systems and interchange btw such systems (version
control systems, e.g.,) - needs more specifics about states of the world.
... Understanding that these two were clashing will help us move beyond.
... Address first use case first.
...Second: "proper" provenance. We are simplifying. In month or so, something nice
to read.
... Kai Eckert, as part of group, working on BP on using Dublin Core with
provenance, so know how this works with model.
... Ivan was there.
Ivan: One more result. At moment, docs structured so
that if someone wants to understand what is happening, read data model which uses
abstract syntax, only then can you move on to OWL and begin to use in SW setting.
... Would start reading with Primer -> Data Model -> OWL: not really helpful
for LD community.
... Future docs will be edited so that Primer -> OWL right away. Goal to have
this order to LD people.
... Also (more image than content): OWL, it turns out, is very simple - manageable
by LD community, but this fact is well-hidden in document.
... A major feature, which messaging will bring out.
Paul++ re: simplification
Ivan: Issue that comes up: Named Graphs. Document can do hand-waving for now.
David: RDF WG has made progress on NGs. Pat Hayes sees
a way forward acceptable - action to compose something.
... As soon as we know, will feed back. Remains thorn in our side.
Paul: We had a concept around "accounts" - a
complicated feature. Decided to dump alot of the verbiage - focus on where we got
provenance asserttions.
... Can move forward without _dependence_ on NG - because less central to model.
<sandro> GLD must recommend several vocabs, yes.
Ivan: What does CR mean? To implement?
Paul: In case of Provenance, used discussion with SKOS
as a model. Some idea that that was too lightweight.
... Ended up coming up with notion - using SKOS model. We have X number of
implementations which use constructs Y and Z.
... Starting point. Then: have at least two applications implementations that
would exchange provenance information and do some sort of query answering over
that provenance information. Higher hurdle: need two apps that work together.
... Aiming at minimum of two.
...Finally: Simple queries that you can do on model and test cases.
... Take test data and answer simple queries.
... To show that we have an interoperable spec.
Sandro: GLD has a bunch of vocabs in charter.
... e.g., vocab about people. Whether new, or old (e.g., FOAF).
... Largely baffled how we can do CR here. W3C way: interoperate with a test
suite. How to do for vocabulary?
Sandro: Can do for an application.
... Not sure how to test - make sure that code is right.
Paul: Instead of making test cases - simple queries
that one should be able to answer. Simple transitive closure. "How many activities
represented?"
... If you have enough apps that can implement queries - enough.
David: Suggest, in relation to vocabs, if test suite consists of sample instance data - combination of instance data and vocabulary can be thrown to a reasoner for validation of logical consistency. Can perform a lot of operations (transitive closure, eg). Should work well.
Sandro: You'd need operations expected. What would a FOAF system need to implement, eg?
David: "If I have a FOAF person, want to determine if two separate people"
Sandro: That's just OWL reasoner.
David: Yes, a lot of cases like that.
Sandro: How do we get one DB of people to merge with another?
<pgroth> really?
Sandro: Need to know if this vocab will work for that.
<pgroth> that's a huge ask
Ivan: Not necessarily merge, but I as user may want to extract or infer info from dataset that uses FOAF?
Sandro: Underlying use case: I want to query across 100 DBs
<pgroth> you can't
Sandro: What test case could I construct? How do we assure the world? What systems implemented? Tested provided?
<bhyland> In the GLD LOD world, we haven't focused on collecting use cases ...
David: Want to be able to show that for a sample system, sufficient system to disambiguate. In alot of real-world system, can't. But that is not a failure of spec, but because identifying people by non-unique identifiers.
Bernadette: Would put emphasis differently. People in
this and many WGs focused on guidance to make reasoners work, making semantics
work in products.
... In GLD WG, consumers of BPs: looking for more "informative" than normative.
Not building products. Coming from world of data warehouses, XML...
... If they go down path of RDF standards, what do they get?
...Europe: more multilingual issues.
... Haven't gotten to granularity of defining use cases. Personal information is
not big issue here. A lot is geospatial. Regulatory, laws.
... Now trying to get them to put toe in water. Frustrated that we are providing
context and history - "pragmatic provenance". Don't tell me about steps down the
road. Rather: 10 basic things.
Paul: Don't see problem in taking adoption road to interoperability for these vocabularies. Don't think use case approach is workable.
<bhyland> +1 to what Paul is saying!
Paul: Not show interoperability by [solving] use case. Bernadette: take adoption approach. May recommend XYZ vocabs. We will recommend provenance vocabulary.
<pgroth> @bhyland - I hope the GLD group will be using the prov vocabulary
Ivan: Want to ask: When CR phase comes and you have more precise spec of criteria, would like to see well-documented outside of group (e.g., CG wiki) so that groups in future will be helped - clearer idea what to do about vocabulary CRs.
<pgroth> @bhyland - if we need to do anything in the prov group to facilitate what you want, please let us know
Ivan: Want to avoid this uncertainty in future. Document what happens for the CG.
<bhyland> @pgroth, I'll review it again & make sure I raise at tomorrow's GLD WG telecon because to be honest, it wasn't discussed during the recent F2F.
<bhyland> We focused on People (DERI), org (epimorphics) and Cube vocab during the F2F at the end of January. So thank you Paul for raising this so it's given due consideration.
<pgroth> @bhyland - i just heard "pragmatic provenance" so wanted to make sure we had overlap there
Ivan: TPAC meetings in past few years had little SW presence. Good to plan for F2F of your groups in Lyon in November.
<sandro> http://www.w3.org/2012/10/TPAC/ 29 October - 2 November 2012 Lyon, France
Ivan: If we have new Linked Data Pattern group, can plan F2F.
David: RDF WG has discussed not having F2F in that timeframe, but may, because not done. Current economic environment has not allowed us to have everyone in one place. Unclear how well we will be able to manage F2F. But good to get representation in Lyon.
<bhyland> @ivan, thanks for the heads up & recommendation so people can plan well in advance.
David: Agenda request: SemTech in Berlin had wikimedia announcement. Discussion on net how that might impact wikipedia and LD community.
<pgroth> wikidata is driven by Semweb people :-)
Ivan: There will be a discussion on Semantic Web gang on Friday with Danny on call.
<sandro> http://www.w3.org/2011/gld/wiki/F2F2
<sandro> (28 people, day 1)
Bernadette: Had F2F - 12 people in DC and 14 in Deri.
Facilities good! Problem in GLD WG, getting action. Lots of IEs with less
experience.
... In public sector, less urgency.
... F2F good at building bonds. Want to get something out on vocabs and BPs rather
than dribble out slowly. Existing content needs to be pulled together.
... DERI people vocab... Authors of editors' draft. Will be more more informative
than normative guidance.
<pgroth> surprised not to see QUDT (http://www.qudt.org/)
Preliminary feedback: "helpful and practical". Using a Void description... the process... demystifying.
<ivan> QUDT: Quantities, Units, Dimensions and Data Types in OWL and XML
<pgroth> we've seen the need for this
<sandro> this was presented at GLD F2F1, as I recall, as something TQ wanted GLD to do.
Bern: QUDT very relevant. Dealing across countries and metric/empirial. Gist: splash by end of February.
<pgroth> @bhyland for provenance see the Prov-Primer http://dvcs.w3.org/hg/prov/raw-file/default/primer/Primer.html
<bhyland> @pgroth
<bhyland> @pgroth, thanks very much!
Sandro: Agree.
Paul: Especially vocab list, recommended vocabs is
super-important, even outside of govt - eg, pharmaceutical. Keen to see list.
... Also keen on making sure Provenance vocab addresses your use cases. Need
feedback before last call.
<pgroth> we need best practice on exact vocabularies
Bern: Will do. Re: vocabs: fifteen or so common vocabs, want to include list. Others think: not list, but criteria for judging whether vocabs appropriate to be used. Qualitative criteria: about organizational commitment eg. Could it go away? Agree there should be criteria.
<tbaker> LC http://www.loc.gov/marc/transition/news/framework-051311
tbaker: In library world, many interesting developments including announcement by Library of Congress that they're planning to replace MARC with a bibliographic framework, based on Linked Data principles.
Pressure on IFLA to publish their vocabs in RDF.
… Harvard, Stanford, Europeana, British Library also have projects happening in this space. Lots of interest.
… Bibliographic Framework Initiative at LOC want to replace MARC with RDF family of standards.
<tbaker> http://dcevents.dublincore.org/
… Tom explained the history & people involved in this progress. Slow progress initially but picking up now. 5 year anniversary of this effort will happen in April 2012 in London.
<tbaker> http://wiki.dublincore.org/index.php/Full_announcement
… A lot of traditional library standards, with longer older historical roots, are being translated to RDF. These older standards include RDA, FRBR (?), ISBD, others.
… These translations are being made by traditional library standardization groups as opposed to sem web savvy experts which causes some concern by the sem web / library community.
… Concern around these translations & how they'll interoperate in the open world on the Web.
… If standards that are used to express library data play well with Linked Data, it will be *huge* contribution to the LOD cloud.
… Note, this involves politics in the traditional library world. We'll looking to start with a workshop to bring together semantic experts with representatives from more traditional library standards community to find common ground, define broader consumers of this important data.
@tbaker, I'll follow up by email as I notice the same phenomena happening with vocabs and modeling done by traditional modeling people who believe automatic translation to RDF is all the same.
thanks all!