<trackbot> Date: 04 May 2010

Admin

<mhausenblas> PROPOSAL: accept minutes from last telecon http://www.w3.org/2010/04/27-rdb2rdf-minutes.html

<soeren> +1

<juansequeda> +1

<mhausenblas> RESOLUTION: WG accepted minutes from last telecon http://www.w3.org/2010/04/27-rdb2rdf-minutes.html

<mhausenblas> scribenick: nunolopes

<cygri> for next time

Goal of the WG re mapping

<juansequeda> http://is.gd/bTNFm

<mhausenblas> http://docs.google.com/drawings/pub?id=1de31u5wBUheygJTXbxXE7e7H7QI_is9hEXi5dOO5knE&w=960&h=720

mhausenblas: new picture from juansequeda

juansequeda: the picture describes our two options

… option 1 uses the putative ontology

… option 2 uses the domain ontology

… and the result will be subset (non-isomorphic)

… with the relational database schema

… hopefully this image will give us a clear picture of where we are

Ahmed: we have been talking about the second option since before MacTed joined

Orri: I agree with Ahmed that we should support more "expressive"? translations

hhalpin: is there a vocabulary problem with the group?

MacTed: I am not saying that option 2 should dissapear

… option 2 actually is included in option 1

<juansequeda> There were some typos. I just corrected the image: http://docs.google.com/drawings/edit?id=1de31u5wBUheygJTXbxXE7e7H7QI_is9hEXi5dOO5knE

… where this step is visible from the tool or not it is still required

<Ahmed> q

<juansequeda> Apologies, the link is http://docs.google.com/drawings/pub?id=1de31u5wBUheygJTXbxXE7e7H7QI_is9hEXi5dOO5knE&w=960&h=720

… iterating/separating the steps all is satisfied

… the local ontology mapping should not be disregarded

<LeeF> +1 to Ashok's observation

<mhausenblas> +1

Ashok: Anyone disagrees that what we are to standardize is option 2, with option 1 being a subset of option 2?

<soeren> +1

<cygri> +1 to ashok

Ahmed: I agree with ashok.

<harry2halpin> but basically, I think we should do option 2.

… nobody said that specifing the domiain ontology is static

<harry2halpin> it's just that the vocabulary in the use-case document needs to be neutral about *how* to do it, i.e. neutral between ericP's sparql construct approach or a SQL view approach

<mhausenblas> s/...?/agree

<Zakim> LeeF, you wanted to ask what explciit inclusion of the local ontology stuff adds to the specification

LeeF: I agree with Ashok

MacTed do you say that there is some value in explicitly doing the transformation from the local ontology to the domain ontology

… what is that value?

<harry2halpin> no-one is saying that a central isomorphic transformation will not be part of the spec

MacTed: the simple transformation is a vital part of the mapping

LeeF: if the final specification requires a more complex transformation such that the direct mapping is a specific case of it why is it still needed to call it directly

harry2halpin: I would like that the direct mapping should be part of the use case document

… but not necessary

MacTed: yes, tools may hide this direct mapping

<harry2halpin> but that R2ML should allow this direct mapping.

Ahmed: we should select option 2

… are you saying that we should do option 2?

MacTed: I belive option 2 contains option 1

<LeeF> Sounds like everyone is in agreement thta Option 2 contains Option 1

… and that option 1 should not be made invisible

<mhausenblas> yes, indeed, we seem to agree

<harry2halpin> We agree with that MacTed, R2ML should be able to do a local ontology mapping.

<mhausenblas> I'd propose to move to the UCR document now

<LeeF> I think what MacTed is asking for is that the specification explicitly calls out 1 particular transformation: the direct mapping transform

Ahmed: we should have a mechanism that allows the user not only to use the local ontology but also the domain ontology mapping

use case document

<mhausenblas> PROPOSAL send out http://www.w3.org/2001/sw/rdb2rdf/use-cases/ as FPWD

mhausenblas: In the current stage, do you have any objections (specifically for MacTed)

MacTed: yes, there are some changes

… I would like to express the following cases

<harry2halpin> Note that Juan has added texts to make it more use-case like MacTed.

… it still feels more like case studies than a use cases document

mhausenblas: can you write a mail with some concrete changes?

MacTed: I can do that

<MacTed> tthibodeau

<mhausenblas> ACTION: tthibodeau to write up changes for UCR document and send to the list [recorded in http://www.w3.org/2010/05/04-rdb2rdf-minutes.html#action01]

<trackbot> Sorry, couldn't find user - tthibodeau

<harry2halpin> Note that this is a problem with your AC Rep.

<harry2halpin> So MacTed, please have your AC rep approve of your membership in the group!

<soeren> +q

<harry2halpin> There is nothing W3C staff can do until your AC Rep approves.

Ashok: publishing the document does not mean we cannot change it

… it's a first public draft

<mhausenblas> ACTION: mhausenb to take MacTed's changes into account before issuing FPWD [recorded in http://www.w3.org/2010/05/04-rdb2rdf-minutes.html#action02]

<trackbot> Created ACTION-52 - Take MacTed's changes into account before issuing FPWD [on Michael Hausenblas - due 2010-05-11].

… we can further work on it

mhausenblas: yes, we just waited for a few changes

soeren: I feel that uc5 and uc6 are not use-cases but requirements

… the labeling 'functional' and non-functional is also a problem

<mhausenblas> PROPOSAL" move UC5 and UC6 to reqs

<LeeF> Note that UC5 and UC6 are my use cases - I mainly care that they be included as requirements

LeeF: in the end I care that they are accepted as requirements

… but it doesn't make sense to have requirements that don't derive from use-cases

<soeren> +q

harry2halpin: the functional and non functional seem ok

<LeeF> I've always used (for better or for worse) http://www.w3.org/TR/rdf-dawg-uc/#uc as my guide for UC&R gathering

soeren: reg uc5 and 6 are more requirements since all the other use cases are regarding a specific domain

… these are more domain independent

Ahmed: the uc6 is not clear to me what is the use case

LeeF: I can clarify but need to go

mhausenblas: let's sort it out via mail

MacTed: the closest thing to a use case is 5 and 6 together

Ahmed: when I read n6 I don't see that

<LeeF> (FWIW, whoever it was who expressed UC6 as dealing with conflations of shema data and instance data had it exactly right - the idea is that we see many DB tables where some sort of indicator value in the row denotes the specific type of instance that the row represents - we need to be able to map that to an appropriate rdf:type when mapping from the RDB data to the RDF data)

… this should be sent for formal feedback soon

… after that the editors can just do some changes

<harry2halpin> +1 to be made public

MacTed: if this is only a public draft I'm ok with it

Ahmed: it should be more than that

<harry2halpin> Releasing a document for the sake of releasing a document is W3C process.

… we spent a lot of time to release it

MacTed: we should have the document released but with what we agree on

Ahmed: the document is a milestone, we should finalise and send it for review, 1 week feedback

harry2halpin: it's normal W3C process to release working drafts

… we should release it asap

… a brief discussion on some notations should be in order

<soeren> http://en.wikipedia.org/wiki/Non-functional_requirement

mhausenblas: regarding renaming of function requirements to core requirements and non functional to auxiliary

<mhausenblas> PROPOSAL: change func req to core and non-functional to auxiliary

Ahmed: let's see what other people say but I disagree with the change

… let's send the document and wait for comments

mhausenblas: in that case the version 1.24 will be for one week under working group review?

… what will be the process?

Ahmed: let's set some time to reconcile the emails as a group

<mhausenblas> PROPOSAL: send out current version 1.24 for WG internal review, 1 week, feedback via mailing list

<ericP> second

juansequeda: only for internal review

mhausenblas: yes

<juansequeda> +1

<harry2halpin> +1, but then we really have to get consensus on it next week to publish.

mhausenblas: please make sure that you do send a mail to the list so we have a record to work on next week

<mhausenblas> yes, harry2halpin

<harry2halpin> +1 ericP, let's discuss this now.

ericP: juansequeda was making a case for direct mapping with no remaning would be the putative ontology?

juansequeda: 3.1.1 direct

ACCEPTED: send out current version 1.24 for WG internal review, 1 week, feedback via mailing list

juansequeda: does everyone understand 3.1.1 and 3.1.2?

<harry2halpin> it was simple enough for me to understand, but I'm more comfortable with using graph/isomorphism terminology than domain/putative ontology talk.

RESOLUTION: the current version of the UCR at  http://www.w3.org/2001/sw/rdb2rdf/use-cases/ (v1.24) is under review for 1 week, WG internally. All members should review and send in comments via the RDB2RDF mailing list.

MacTed: there seems to be a slight inconsistency in the headings

… 3.1.1 isomorphic

… 3.1.2 non isomorphic

<mhausenblas> MacTed: please note that our resolution somehow has superseded our actions - will remove them again

… 3.1.2 a is currrently 3.1.3 (SQLTran)

juansequeda: so 3.1.4 would be part of the direct

MacTed: yes, think so

… I think label generation comes from the schema and is part of the direct translation

… label transformation doesn't seem to be described

harry2halpin: a direct transformation from the relational schema to a graph

… which terminology should we use?

<harry2halpin> domain/putative ontology?

… domain/putative?

Ahmed: local and domain ontology are well understood outside the RDF comunity

<harry2halpin> so we need to make an edit to the spec to use domain/putative ontology rather than purely direct/non-direct transform.

… I suggest we use those

<Zakim> ericP, you wanted to clarify SQLGEN

ericP: this literature comes from 1992

… not necessarially reg RDF

<MacTed> the beauty of this is we can define the term in the document.

<harry2halpin> we can then specify what precisely we mean in terms of RDF.

mhausenblas: I also suggest local and domain ontology

ericP: where dumping the data into a graph that can be described by an ontology

harry2halpin: we should use a vocabulary widely used like from the database comunity

… but we can use the terms if we define them properly

juansequeda: we translate the data which is an instance of the ontology

… option 1 uses a putative/local ontology

… option 2 uses a domain ontology

<mhausenblas> PROPOSAL: use local and domain ontology throughout the document and make clear that the data is an instance

ericP: we should be able to run the transform in both ways.

… take a sparql query and turn it into a SQL query

MacTed: the transformation should be reversible?

ericP: yes

<harry2halpin> +1

<juansequeda> +1

RESOLUTION: use local and domain ontology throughout the document and make clear that the data is an instance

<harry2halpin> "Coercing the relational graph into this pattern requires graph transformation."

<harry2halpin> dump to RDF->then use SPARQL constructs that direct RDF graph.

<harry2halpin> coercing the relational data into this pattern requires non-isomorphic graph transformations, i.e. transformation into a domain ontology

<harry2halpin> "coercing the relational data into this pattern requires non-isomorphic graph transformations, i.e. transformation into a domain ontology"

<mhausenblas> [adjourned]

<ericP> ACTION: ericP to send a READ ME request [recorded in http://www.w3.org/2010/05/04-rdb2rdf-minutes.html#action03]

<trackbot> Sorry, couldn't find user - ericP

<ericP> ACTION: eric to send a READ ME request [recorded in http://www.w3.org/2010/05/04-rdb2rdf-minutes.html#action04]

<trackbot> Created ACTION-53 - Send a READ ME request [on Eric Prud'hommeaux - due 2010-05-11].

This is scribe.perl Revision: 1.135  of Date: 2009/03/02 03:52:20  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

FAILED: s/...?/agree/
Succeeded: s/…?./agree/
Succeeded: s/… MacTed/MacTed/
Succeeded: s/??/LeeF/
Succeeded: s/MacTed/mhausenblas/
Succeeded: s/harry2halpin/MacTed/
Succeeded: s/ausiliary/auxiliary/
Succeeded: s/?/harry2halpin/
Found ScribeNick: nunolopes
Inferring Scribes: nunolopes
Default Present: whalb, juansequeda, MacTed, mhausenblas, cygri_, +39.046.128.aaaa, nunolopes, cygri, soeren, Ashok_Malhotra, hhalpin, Lee_Feigenbaum, harry2halpin, EricP
Present: whalb juansequeda MacTed mhausenblas cygri_ +39.046.128.aaaa nunolopes cygri soeren Ashok_Malhotra hhalpin Lee_Feigenbaum harry2halpin EricP
Regrets: Marcelo Seema Souri Dan
Agenda: http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2010May/0002.html
Found Date: 04 May 2010
Guessing minutes URL: http://www.w3.org/2010/05/04-rdb2rdf-minutes.html
People with action items: eric ericp mhausenb tthibodeau

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.