RDB2RDF Working Group Teleconference -- 12 Apr 2011

<trackbot> Date: 12 April 2011

<ivan> ok

<ivan> And I have already scribed today at another meeting...

<mhausenblas> ericP?

<mhausenblas> juansequeda, in case ericP can't scribe, would you be able to?

I was expecting you to ask me that :)

I guess I haven't scribed in a while

<mhausenblas> scribenick: juansequeda

Admin

<ivan> do you still have the problem?

<mhausenblas> no

<mhausenblas> PROPOSAL: Accept the minutes of last meeting http://www.w3.org/2011/03/29-rdb2rdf-minutes.html

RESOLUTION: the minutes from last meeting have been accepted

Towards REC

<Souri> 007 => James Bond

<mhausenblas> http://www.w3.org/2005/10/Process-20051014/tr

mhausenblas: Ivan will describe what the next steps are for formal requirements
... based on http://www.w3.org/2005/10/Process-20051014/tr
... which captures all the W3C process

ivan: we have had several drafts which is fine
... and it is up to us to decide how many of those to publish
... we have to come to a Last Call Working Draft
... and this means that the WG thinks that it has solved all the technical issues

<mhausenblas> Michael: LC see http://www.w3.org/2005/10/Process-20051014/tr#last-call

ivan: there shouldn't be any open issues
... and we should get their soon
... the Last Call Working Draft should lead to comments from the community
... and if we are lucky, we won't get any comments
... if we do get comments, then we have to reissue a new Last Call Working Draft
... if we are lucky/we did a good job, the community should be happy with the Last Call Working Draft
... we have to tell the community that they have 6 weeks.. 2 months to give comments back
... and we need to document how we address the community comments
... after the Last Call Working Draft comes the Candidate Recommendation
... this is to prove that the technology can be implemented
... during this phase, we call for implementations
... we expect 2 mutually independent implementations
... we need to produce a load of test cases
... we hope to have several implementations
... we need a way to document. implementation x has features a,b,c
... revelytix has an implementation of R2RML, alex/eric has a direct mapping, juan has a direct mapping
... we need another R2RML implementation
... the candidate recommendation phase can show two options
... 1) one of the features is almost impossible to implement
... and if this is the case, then we have to go back to Working Draft
... 2) we did a good a job and no problems should arise
... after Candidate Recommendation, we go to Proposed Recommendation
... and the the AC members of the W3C will vote on this document in order for it to become a recommendation
... and that is the process
... There is a lot of admin stuff that needs to be done (documenting what we did, how we addressed the community comments, etc)

mhausenblas: we need to document all the comments even if they are contradict eachother
... what happens if there is a formal objection?

<Zakim> mhausenblas, you wanted to note on handling issues/formal objection

ivan: we need to avoid formal objections
... and we need to find a consensus
... and that is why it takes so long to come up with a standard
... there are cases where we don't come to a consensus on a formal objection
... and if so, it goes up to the director (timbl), and it is his decision
... however, we have never had a case like this

ericP

;

ericP: we need to show which issues came from outside

ivan: ... (talking about how rdfa and owl WG have organized themselves in order to address community comments)

mhausenblas: souri, are you guys planning to implement this?

souri: we are planning to implement R2RML and this will coincide with a new release of oracle

ivan: in the candidate recommendation phase, the implementation should be product quality

souri: ... beta release is a limited release

ivan: this is fine for a candidate recommendation phase
... if you state that you have implemented x amount of features, that is fine

souri: we already had an implementation of the direct mapping, even before the spec came out

ericP: does your direct mapping produce an RDF graph or can you only query it with SPARQL

souri: you can query it with SPARQL
... we get a SPARQL query and translate to a SQL query

ericP: can you then do ASK?

souri: yes

ivan: in the OWL group, Oracle appears in the candidate recommendation phase, and this was before the official release of oracle

souri: yes, this was implemented before

mhausenblas: we need to organize the timeframe now
... same question to Openlink

MacTed: I have no information on dates

mhausenblas: could you figure it out?

MacTed: ok

mhausenblas: david, any critical dates?

dmcneil: we do not have plans to make a direct mapping implementation

soeren: working on something. but we don't have concrete dates.

juansequeda: we will have an implementation of direct mapping ready for June

betehess: eric/alex will be presenting at K-CAP

<betehess> http://kcap11.stanford.edu/

juansequeda: marcelo will be giving a tutorial at SIGMOD/PODS and juan will be at the semantic web summer school talking about rdb2rdf

<betehess> would be good to have something close to LC (at least very stable) because of all the conferences

mhausenblas: we need more exposure in the industry side

souri: we will be doing something at Oracle Open World

ericP: i was at Enterprise Data world and I talked about it on a panel that I was at

<mhausenblas> http://www.w3.org/2001/sw/rdb2rdf/wiki/RDB2RDF_Publicity

<Zakim> betehess, you wanted to speak about exposure in industry

ivan: to make clear, for the candidate recommendation phase, there is no requirement, that the implementation has to come from the working group
... if there is any organization outside.. even better

betehess: what are the concrete next steps in order to get to the Last Call

ivan: we still have open issues. these need to be closed

Test Harness

<alexdeleon> http://www.w3.org/2001/sw/rdb2rdf/wiki/TestHarness

<mhausenblas> http://www.w3.org/2001/sw/rdb2rdf/wiki/TestHarness

alexdeleon: here we have 3 proposals, which I think the third one is the most adequate
... we need to publish the test cases in a machine consumable way
... each test has its own URI which is resolvable
... maybe optional report run time
... these are for materialized RDF

<mhausenblas> Michael: the test cases manifest is in fact http://www.w3.org/2001/sw/rdb2rdf/test-cases/ - in RDFa ;)

alexdeleon: for SPARQL to SQL implementations, the testing is a bit more complicated

<mhausenblas> http://www.w3.org/2007/08/pyRdfa/extract?uri=http://www.w3.org/2001/sw/rdb2rdf/test-cases/

mhausenblas: the current test-cases are in RDFa

sorry.. I'm a bit lost on scribing.

<alexdeleon> http://www.w3.org/2001/sw/rdb2rdf/test-cases/

mhausenblas and alexdeleon talking about machine consumable test cases

alexdeleon: for posting the results of the test?

<mhausenblas> http://sparql.org/sparql.html

<Zakim> mhausenblas, you wanted to comment on manifest

<betehess> +1

ericP: we need to test for exact graph match.
... we can make a CONSTRUCT query
... and this will give you a sparql report which is the full RDF graph
... and then you can compare two graphs

<betehess> actually, exact graph match as graph isomorphism

ericP: we are dealing with small graphs

<Zakim> betehess, you wanted to ask if we should introduce a dependency from the Test Harness to both Direct Mapping and R2RML? (/me would say yes)

betehess: should both direct mapping and R2RML document reference the Test Harness
... we should publish three documents at once
... and reference the Test Harness

<mhausenblas> Michael: not the Test Harness, but the Test Cases http://www.w3.org/2001/sw/rdb2rdf/test-cases/

<ericP> http://www.w3.org/2001/sw/DataAccess/tests/r2

<ivan> the OWL WG just had a test cases on a separate wiki site, actually...

<ericP> http://www.w3.org/2001/sw/DataAccess/tests/README.html

betehess: we don't really care about versioning for the test suite?

ericP: things were volatile for a while, and then we froze it

mhausenblas: anything else on this topic

<ivan> bye guys

<mhausenblas> [meeting adjourned]

<mhausenblas> trackbot, end telecon

- DRAFT -

RDB2RDF Working Group Teleconference

12 Apr 2011

Attendees

Contents

Admin

Towards REC

Test Harness

Summary of Action Items

Scribe.perl diagnostic output