RDB2RDF Working Group Teleconference -- 09 Nov 2010

<trackbot> Date: 09 November 2010

<mhausenblas> hey Ashok - will you be on the phone as well or IRC-only?

<MacTed> hm.

<MacTed> drats.

<mhausenblas> scribenick: betehess

<mhausenblas> http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2010Nov/0047.html

mhausenblas, first we go to the minutes of the last meeting

admin

<mhausenblas> PROPOSAL: http://www.w3.org/2010/10/26-rdb2rdf-minutes.html

mhausenblas, any comment?

<hhalpin> +1

mhausenblas, minutes approved

default mapping

mhausenblas, I sent the links to the two documents yesterday

mhausenblas, tlr is gonna talk there

tlr: thanks for letting me joining the calk
... RDB2RDF if part of my domain
... I'm glad we have the WD for r2rml
... and it's great that people work together (seeing the ML)
... but I'm concern about the default mapping
... I hope there will be a way to see this default mapping
... thanks Juan, Marcelo and Eric about that

mhausenblas: we just delayed a bit the default mapping to focus on r2rml before TPAC
... now, I'd like the 2 editing teams to present their work
... please note the document MUST be *complete*

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

mhausenblas: richard contributed quite a lot to this one
... unfortunately he is not around
... what's the status there?

ericP: I've offered to merge in this document

mhausenblas: it's your document, without Juan and Marcelo bits?

ericP: yes
... it follows the typical W3C structure
... chapt 1 is an introduction
... chap 2 is an informative section
... introduces about what's happening
... the normative section would be less friendly
... there is a bit of DDL, describing the database
... then you see the Direct Graph, resulting from the mapping
... I have to remove a section
... then I explain how to generate triple from a SQL row
... I believe it's good to provide URIs ready for Linked Data
... even if it wasn't explicitly required
... then I go into tricky cases
... involving FK, PK, etc.
... then about structural table
... I try to model the Direct Graph so it reflect the hierachical table

<Souri> You can have emp/manager relationship, where fkey=pkey (self-loop)

ericP: but someone told me it wasn't totally good
... that's an issue I have to solve
... what if the FK has the same col as the PK but in a different order?

Souri: you're trying to deal with normalized table, right?
... so we would like to have the same URI here

ericP: yes

Souri: all the groups of tables needs the same URI as a subject

MacTed: all that stuff is beyond the Direct Mapping

ericP: another issue with many-to-many
... it's documented and is discussed on the ML
... then I use a set notation
... remark: if you go over cells in the table, you'll see the relation with generated triple
... section 3, I use boring set notations
... then the definitions for the mapping
... next section is non normative (copy the RDF spec)
... then I explain how to go from RDB to RDF

<mhausenblas> http://www.w3.org/2001/sw/rdb2rdf/directGraph/alt

mhausenblas: then Juan and Marcelo

juansequeda: we have a similar intro
... section 2 is one the biggest difference
... the mapping is in 2 sub sections
... first the URIs and then the Triple
... because we think it's complicated
... we don't have strong feelings about the notations
... tried to answer richard's comments

<hhalpin> Notes there are minor diffs between IRI generation in Eric and J&M document, but these can easily be reconciled - likely the result of text changing in one draft and not another.

juansequeda: we compile the exceptions inside this rule

<hhalpin> i.e. a reason to converge drafts :)

juansequeda: in 2.3
... all the rules are compile into Triple(...)
... section 3
... this is where we define the semantics
... using datalog rules
... we have to define where the section 3 should be
... we want to keep the Default Mapping as simple as possible

mhausenblas: we can put that in notes when we'll have one document
... thanks again to the editors
... it seems the diff is not too big

<ericP> i did get feedback from mike stonebraker and he also said he prefers to not see many-to-many

<mhausenblas> http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2010Nov/0052.html

<mhausenblas> http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2010Nov/0057.html

mhausenblas: (based on the emails above)

<Ashok> We can create issues for the M-M mapping and hierarchical tables

juansequeda: +1 for for easy merging

Ashok: is it Eric's + datalog rules?

juansequeda: we want to generate the generation of URIs and then the mapping rules

Ashok: it's a good idea

<hhalpin> go for souri first, i'm muted

<juansequeda> hello harry?

<hhalpin> ah, unmuted now.

Souri: I found J&M's section 2 easy to read

hhalpin: we really need one document
... ericP did a good job, especially on the example
... but found the URI generation easier to follow
... no single individual should ever edit a document by himself

mhausenblas: let's create a new document
... copying the content

<mhausenblas> PROPSAL: http://www.w3.org/2001/sw/rdb2rdf/defaultMapping/ with Juan, Marcelo and Eric as editors based on Richard's propsal

ericP: 2 things
... I should add Alexandre as an editor for the formalism
... I'd propose to call it Direct Mapping

<mhausenblas> PROPSAL: http://www.w3.org/2001/sw/rdb2rdf/defaultMapping/ with Juan, Marcelo and Eric as editors based on Richard's proposal and have the IRI section explicit

<MacTed> +1 "directMapping" (not defaultMapping)

ericP: especially cause we speak about the Direct Graph being the result of the Direct Mapping

<MacTed> +1 for proposal otherwise

ericP: to avoid confusion with Default Graph in r2rml

<mhausenblas> PROPSAL: http://www.w3.org/2001/sw/rdb2rdf/directMapping/ with Juan, Marcelo and Eric as editors based on Richard's proposal and have the IRI section explicit

<ericP> http://www.w3.org/2001/sw/rdb2rdf/directGraph/#iris

ericP: here is the section about how I generate IRIs

mhausenblas: ericP added that recently based on feedbacks

<hhalpin> Let's put hierarchical tables as an open issue

<hhalpin> there is clearly disagreement in the WG over this example.

<mhausenblas> PROPOSAL: http://www.w3.org/2001/sw/rdb2rdf/directMapping/ with Juan, Marcelo and Eric as editors based on Richard's proposal as of http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2010Nov/0052.html and have the IRI section explicit, move hierarchical table and the M-M mappings into Ed note

mhausenblas: the WG has to discuss this issue

Ashok: the document will have the datalogs rules

<hhalpin> I think as an option.

Ashok: per richard's proposal

<mhausenblas> Michael: yes, we should create issues for both hierarchical table and M-M mappings

mhausenblas: so we add it in the document as an issue

<hhalpin> key word "can live with" :)

juansequeda: something has to be changed
... in 2.2
... @@

<ericP> http://www.w3.org/2001/sw/rdb2rdf/directGraph/#reference_triple ?

<MacTed> it seems we can take it as given that "the overall document must be edited to make the merged sections fit/make sense"?

hhalpin: I would like a consensus on the text, not formalisms
... we're not talking about adding datalog but more the description of IRIs sections
... I believe it's clearer

<Zakim> ericP, you wanted to address iris section

<ericP> http://www.w3.org/2001/sw/rdb2rdf/directGraph/#iris

ericP: I believe I *have* the IRIs section

<hhalpin> to clarify, I'm saying that 2.2 is as is difficult to read, and I think Juan and Marcelo's section 2.2 and 2.3

ericP: which richard proposed to re;ove

mhausenblas: it doesn't harmed to have this ection
... then we can ask the community later

<tlr> I think co-author is the thing you want, Eric.

hhalpin: I leave the question of adding Alexandre in the hands of the chairs

<hhalpin> and we generally want more than one editor as editors often flake :)

ericP: traditionally, someone brings something and the the chairs chose

<ericP> +1

Ashok: what about the datalog rules?

mhausenblas: does it impact the direct mapping?

Ashok: it doesn't
... I just would like it to be present

mhausenblas: I propose that after the call, the 3 editors make the merge
... with the goal to published next week

<MacTed> PROPOSAL: http://www.w3.org/2001/sw/rdb2rdf/directMapping/ with Juan, Marcelo and Eric as editors based on Richard's proposal as of http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2010Nov/0052.html and try to work in J&M's IRI and Triple generations part; move hierarchical table and the M-M mappings into Ed note; datalog rules replace scala rules (under button control); Eric perform merge with review/approval/consensus of Juan, Marcelo, & Eric

<MacTed> for final product...

ericP: I just need 2 hours

Ashok: you'll have one document, with issues in it
... after that, we can vote

<ericP> +1

<ericP> (to MacTed's proposal)

<MacTed> Juan, Marcelo, & Eric for final product...

mhausenblas: any objection?

<MacTed> +1

<mhausenblas> +1

<juansequeda> +2

<Marcelo> +1

<hhalpin> +1

<juansequeda> I meant +1

<boris> +1

<hhalpin> I'm happy with two separate sections, or buttons.

ericP: Ashok's proposal is not to add an entire section
... but to put that insterad of scala

Ashok: exactly!
... ericP, let J and M doing it

<mhausenblas> PROPOSAL: http://www.w3.org/2001/sw/rdb2rdf/directMapping/ with Juan, Marcelo and Eric as editors based on Richard's proposal as of http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2010Nov/0052.html and try to work in J&M's IRI and Triple generations part; move hierarchical table and the M-M mappings into Ed note; datalog as a separate section; Eric perform merge with review/approval/consensus of

<juansequeda> +1

<Marcelo> +1

<mhausenblas> +1

<MacTed> +1

<hhalpin> +1

<boris> +1

mhausenblas: hopefully, we have a FPWD next week
... thanks a lot everyone again

<mhausenblas> [adjourned]

<MacTed> RESOLUTION: http://www.w3.org/2001/sw/rdb2rdf/directMapping/ with Juan, Marcelo and Eric as editors based on Richard's proposal as of http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2010Nov/0052.html and try to work in J&M's IRI and Triple generations part; move hierarchical table and the M-M mappings into Ed note; datalog as a separate section; Eric perform merge with review/approval/consensus of Juan, Marcelo, & Eric for

<MacTed> final

<MacTed> :-)

<MacTed> scribenick: MacTed

<betehess> yes?

<hhalpin> scribenick: MacTed

RESOLUTION: http://www.w3.org/2001/sw/rdb2rdf/directMapping/ with Juan, Marcelo and Eric as editors based on Richard's proposal as of http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2010Nov/0052.html and try to work in J&M's IRI and Triple generations part; move hierarchical table and the M-M mappings into Ed note; datalog as a separate section; Eric perform merge with review/approval/consensus of Juan, Marcelo, & Eric for

final

trackbot, end meeting

<mhausenblas> trackbot, get lost

<trackbot> Sorry, mhausenblas, I don't understand 'trackbot, get lost'. Please refer to http://www.w3.org/2005/06/tracker/irc for help

<hhalpin> trackbot, go find a hobby

<trackbot> Sorry, hhalpin, I don't understand 'trackbot, go find a hobby'. Please refer to http://www.w3.org/2005/06/tracker/irc for help

<mhausenblas> trackbot, women

<trackbot> Sorry, mhausenblas, I don't understand 'trackbot, women'. Please refer to http://www.w3.org/2005/06/tracker/irc for help

<mhausenblas> me neither

<hhalpin> +1 trackbot

<mhausenblas> ok, enough fun - thanks guys

RDB2RDF Working Group Teleconference

09 Nov 2010

Attendees

Contents

admin

default mapping

Summary of Action Items