W3C

- DRAFT -

RDB2RDF Working Group Teleconference

15 Feb 2011

See also: IRC log

Attendees

Present
dmcneil, Ivan, Alexandre, EricP, privera, Souri, +1.603.897.aaaa, Seema, cygri, Ashok_Malhotra, +575737aabb, juansequeda
Regrets
Boris, Michael, Ashok, Sören
Chair
Ivan
Scribe
dmcneil, ericP

Contents


<trackbot> Date: 15 February 2011

admin issues

<ivan> scribenick: dmcneil

<ericP> dmcneil: i can scribe

<ericP> ... but i need to know the incantations

admin

ivan: starting meeting with admin

<ivan> last week's minutes: http://www.w3.org/2011/02/08-rdb2rdf-minutes.html

<ericP> i second minutes

<betehess> +1

ivan: meetings are accepted
... minutes are accepted
... test cases are on the agenda, but Boris sent his regrets
... any test case related issues for the call?

<betehess> /me would like to thank Boris for participating to http://this-db-really.does-not-exist.org/

ericP: we can still try out the test cases and see if they make sense

ivan: proposes moving to R2RML issues

R2RML

<Ashok> RESOLUTION: Minutes from last week are accepted

<Souri> We can discuss ISSUE-23 raised by David

<ivan> ISSUE-23?

<trackbot> ISSUE-23 -- Make tableOwner optional -- raised

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/23

<ericP> scribenick: ericP

dmcneil: current R2RML spec requires the tableowner set when using tablename
... propose to make it optional to reflect SQL practice

Souri: discussed with seema and @@1.
... it makes sense for it to be optional
... at query processing time, we'll be connected so we can resolve this via a query time binding

ivan: clear what doc mod is required?

Souri: yep, we just change the cardinality

<cygri> +1 from me

<scribe> scribenick: dmcneil

<Souri> PROPOSE: make rr:tableOwner property optional (that is, have a cardinality of 0 or 1)

<Ashok> Souri: Recommends that the WG accept the proposed solution to Issue-23

<ivan> +1

<cygri> +1

<Souri> +1

<betehess> +1

<ivan> RESOLVED: make rr:tableOwner property optional (that is, have a cardinality of 0 or 1)

cygri: do we need to keep an issue open to include this in testing? (paraphrased by ericP)

Ashok: ISSUE-23 is not open, create an action to apply it

<ivan> ACTION: Souri to implement the conditionality of table ownder [recorded in http://www.w3.org/2011/02/15-rdb2rdf-minutes.html#action01]

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

<Souri> sdas2

<ivan> ACTION: sdas2 to implement the conditionality of table ownder [recorded in http://www.w3.org/2011/02/15-rdb2rdf-minutes.html#action02]

<trackbot> Created ACTION-103 - Implement the conditionality of table ownder [on Souripriya Das - due 2011-02-22].

<ivan> ISSUE-92?

<trackbot> ISSUE-92 does not exist

<ivan> ACTION-92?

<trackbot> ACTION-92 -- Richard Cyganiak to address the R2RML - DM connection -- due 2011-02-08 -- OPEN

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/92

cygri: haven't done it yet, quite a bit of traffic on mailing list, need to think about how to do it in the document
... there was a discussion about whether there should be a primer, or something that explains how the working group deliverables work together
... perhaps this could be addressed in the primer (or overview doc)
... or we could just add a bit of text to each of the docs

ericP: withdraws question regarding whether a connection exists at times of query rewriting

ivan: strongly in favor of an overview/primer document

Ashok: yes, we should write a primer, but people might not read it
... also, add at least one paragraph to each doc talking about the relationships between the docs

<ericP> OWL primer?

ivan: may not agree

<ivan> http://www.w3.org/TR/owl2-overview/

ivan: points to the OWL document that is brief (1.5 pages) that shows how all of the pieces fit together

<juansequeda> +1 to an overview document

ivan: this document was well received

cygri: probably a short document because there is not a long list of deliverables (like OWL), but it will make it seem more complicated
... if it can be addressed easily in the main document, then it is a good idea

cygi: the main documents are Direct Mapping and R2RML documents

cygri: different paragraph in each referring to the other
... this would produce a more usable set of documents
... could try putting the content in the wiki and then assess whether it warrants a separate document

<Zakim> betehess, you wanted to comment quickly on how we could use the spec grouping feature, like http://www.w3.org/standards/techs/rdfrel.html#w3c_all

<betehess> http://www.w3.org/standards/techs/rdfrel.html#w3c_all

betehess: it is possible to create a new group to have all the specification in one group
... dependencies between documents is difficult to manage

Souri: the R2RML document needs to be extended to describe how direct mappings can be used in a custom mapping

<betehess> /me would like to recall we'll also have a document for the tests

Souri: need examples/test cases that show such hybrid mappings, this could go in a "primer"

<betehess> ivan, open an issue :-)

<Souri> illustrating hybrid mappings (some portion R2RML, some portion DM)

ericP: currently the Direct Mapping refers to the R2RML spec
... the R2RML spec will need a reference to the Direct Mapping for the feature of using Direct Mappings in a custom mapping
... so this will be covered anyway

juansequeda: short, summary document could be a good thing for the wider community

ivan: a summary document that clearly shows where to start reading makes sense

<ivan> ACTION-93?

<trackbot> ACTION-93 -- Souripriya Das to identify not-mapping vs. default-mapping issues in R2RML/DM (Wiki or via mail) -- due 2011-02-08 -- OPEN

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/93

ivan: re-interpret ACTION-92 to be write proposed content on the wiki

Souri: discussed general approach last week, haven't worked on it much, not difficult to get done

<ivan> ACTION-97?

<trackbot> ACTION-97 -- Richard Cyganiak to look into D2RQ implementation and update http://www.w3.org/2001/sw/rdb2rdf/wiki/Entity_disambiguation -- due 2011-02-08 -- OPEN

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/97

Souri will write down more details

on Action-93

cygri: ACTION-97 is to document how D2RQ supports translation tables

ivan: why is this relevant?

cygri: charter requires that we try and re-use entity URIs, question is how to get external URIs into the output of the mapping

<ivan> ACTION-96?

<trackbot> ACTION-96 -- Richard Cyganiak to capture this (multiple subject maps) and other (future) potential error cases in the Wiki (ISSUE-16) -- due 2011-02-08 -- OPEN

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/96

<ivan> ISSUE-16?

<trackbot> ISSUE-16 -- Should we allow multiple subjects for a logical table row? Also, what if no subject is specified? -- open

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/16

cygri: ACTION-96 - since R2RML is expressed in RDF and there is not a schema that clearly expresses what is valid, wanted to collect cases where cardinality was violated
... we decided to collect such cases on a wiki page (e.g. "problem multiple subject maps")
... for each of those we could document the solution (e.g. "raise error", etc.)
... then we could use the wiki page contents to drive the specs
... agrees to keeping this action item open

<Souri> We need to prioritize ... and in my opinion, Primer should come later (we can start on it, but), we need to push out the next draft sooner

ivan: the R2RML editor's draft is quite different from the official draft
... when can we produce the next official draft
... the gap between the two drafts is so large that an updated draft needs to be done soon

Souri: agrees, and thinks the official draft can be updated with issues marked
... ISSUE-93 - the constructs for hybrid maps is still not defined, but this can be marked as still "in progress"

cygri: there has been good progress on the Direct Mapping, might be a draft soon, might be able to publish them together?

<betehess> we should fix a date where we publish both documents

<betehess> +1 to ivan :-)

<cygri> +1 to publishing at same time

ivan: very much in favor of publishing the two documents together, let's return to the Direct Mapping in a moment

<betehess> ivan, we should publish the test-suite NOTE

is that juan talking?

juansequeda: will be traveling to Chile soon and will get a large amount of work done then
... on the Direct Mapping doc (?)

<Souri> 15-Mar-2010?

<Souri> 2011

Direct Mapping

ivan: let's not fix a precise date now

<ivan> ACTION-98?

<trackbot> ACTION-98 -- Juan Sequeda to rename Issue 11 and fix it in the DM -- due 2011-02-08 -- OPEN

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/98

ivan: on to the Direct Mapping, 3 open actions

<ivan> ISSUE-11?

<trackbot> ISSUE-11 -- Primary Key is a Foreign Key -- open

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/11

juansequeda: I did ACTION-98 already

juansequesa: this issue was where "foreign key" was incorrectly typed as "candidate key" in the spec

<ivan> ACTION-99?

<trackbot> ACTION-99 -- Juan Sequeda to change ISSUE-13 to postponed and add ref from ISSUE-11 to ISSUE-13 -- due 2011-02-08 -- OPEN

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/99

juansequeda: the issue is still open (re ACTION-98)

<Ashok> Trackbot, Close ACTION-98

<trackbot> ACTION-98 Rename Issue 11 and fix it in the DM closed

<ivan> ACTION-102?

<trackbot> ACTION-102 -- Ted Thibodeau to sum up the possibilities for generating reliable URIs for DM (to avoid bNodes) on the Wiki -- due 2011-02-15 -- OPEN

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/102

ivan: ACTION-102 still open because Ted is not here
... what is the status of the Direct Mapping?

ericP: there is not much new information, new information will come from Juan's work in Chile (with Marcello (?))
... an open question is whether the Direct Mapping produces persistent identifiers for rows without primary keys

juansequeda: there is still an open issue related to avoiding blank nodes
... there are several issues that we can begin discussing: e.g. hierarchical relationships, etc.

<betehess> the denotational semantics does *not* need a primary key and uses bnode very well https://dvcs.w3.org/hg/FeDeRate/file/0407570a9826/directmapping/src/main/scala/DirectMapping.scala#l37

ivan: would like to come back to the publication issue before the end of the call
... what open issues to dicsuss now?

ericP: maybe we can knock off the "primary key is a foreign key" issue

<betehess> you can try hier_table in http://this-db-really.does-not-exist.org/ with and without "Detects hierarchy relation"

ericP: there was a model for the mapping, the considered the case of whether a primary key is also a foreign key, this is one of the 3 ways users express hierarchies in RDBMS, created a solution for expressing this in the mapping
... the behavior with "detects hierarchy" turned off is the simple behavior

<Zakim> juansequeda, you wanted to discuss the three possible cases

ericP: the proposal is that this simple behavior be the behavior that is implemented when a primary key is a foreign key

juansequeda: there are three possibilities: 1) always detect hierarchy 2) never detect as hierarchy 3) make it an option to detect this as a hierarchy

<betehess> ivan, yes

ivan: how would option "3" work, the beauty of the Direct Mapping is it just works

juansequeda: there could be "flavors" of direct mappings

ivan: would rather not see a switch (personal opinion)

<Zakim> betehess, you wanted to say I'm against any switch

betehess: agrees with ivan

<cygri> +1 to have only one flavour. if you want switches, go to r2rml, it has plenty

betehess: put the switch into the demo application to show what is being discussed, not to advocate for a switch in the Direct Mapping

ivan: proposes publishing the document with the simplest approach, put in a statement that this is still an open issue, ask community for feedback

<betehess> sounds like a wise proposal

juansequeda: agrees with ivan

<Zakim> ericP, you wanted to say we've already done one round without feedback

<privera> +1

Ashok: agrees with ivan

ericP: we tried that, and the community did not respond
... so proposes that we just go with the simple solution

ivan: has no problem with that

ericP: the compromise is that most primary keys appear as literals in the output triples

<juansequeda> ericP, which example are you talking about?

ericP: but id the PK is also a FK then it appears only as a reference to another triple

cygri: in general agrees with proposal to proceed, there wasn't a really clear answer, so simplest is safest

<Souri> To me pk=fk case seems to indicate presence of a simple 1:1 join between a parent table and a child table (which often appears in a normalized DB design) ... is there much more to it?

<ericP> PROPOSAL: unary primary keys which are also foreign keys are treated exactly as other unary foreign keys

<betehess> for the record, this is where we don't take unary foreign key in the scalars function: https://dvcs.w3.org/hg/FeDeRate/file/0407570a9826/rdb/src/main/scala/RDB.scala#l129

<ivan> +1

<ericP> +1

<Souri> s/were/where/

<juansequeda> +25

<betehess> +1

+1

<privera> +1

<ivan> RESOLVED: unary primary keys which are also foreign keys are treated exactly as other unary foreign keys

<Souri> 0 (not fully clear)

ivan: can we stay on for a few minutes to discuss publishing Direct Mapping
... proposes that the Direct Mapping document be published at the same time as the R2RML document without the Chilean work from Juan

<betehess> are the r2rml people ready to publish too?

<Zakim> ericP, you wanted to suggest we lose review attention

juansequeda: progress has been made since the current workign draft

ericP: fears publishing document with few changes because it endangers the public review capacity

<Souri> I am ok with publishing R2RML next draft alone

<betehess> +1 to ivan comment re: publication of both documents makes sense

ivan: agrees with respect to Direct Mapping, but R2RML doc has changed a lot, needs to be published soon, is a good habit to publish them together

betehess: agrees with ivan

ericP: thinks the review capacity is more important

ivan: meeting is adjourned

<privera> +1 to member:ivan comment re: publication of both documents makes sense

<cygri> great job dmcneil!

<ericP> PROPOSED: dmcneil is scribe for life

Ashok: will publish minutes

<betehess> bye

<juansequeda> +25

<ericP> resolved

<juansequeda> ericP, I think dmcneil has been the best scribe ever!

Summary of Action Items

[NEW] ACTION: sdas2 to implement the conditionality of table ownder [recorded in http://www.w3.org/2011/02/15-rdb2rdf-minutes.html#action02]
[NEW] ACTION: Souri to implement the conditionality of table ownder [recorded in http://www.w3.org/2011/02/15-rdb2rdf-minutes.html#action01]
 
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2011/02/15 18:05:02 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
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)

Succeeded: s/tuple/triple/
Succeeded: s/is were/is where/
FAILED: s/were/where/
Found ScribeNick: dmcneil
Found ScribeNick: ericP
Found ScribeNick: dmcneil
Inferring Scribes: dmcneil, ericP
Scribes: dmcneil, ericP
ScribeNicks: dmcneil, ericP
Default Present: dmcneil, Ivan, Alexandre, EricP, privera, Souri, +1.603.897.aaaa, Seema, cygri, Ashok_Malhotra, +575737aabb, juansequeda
Present: dmcneil Ivan Alexandre EricP privera Souri +1.603.897.aaaa Seema cygri Ashok_Malhotra +575737aabb juansequeda
Regrets: Boris Michael Ashok Sören
Found Date: 15 Feb 2011
Guessing minutes URL: http://www.w3.org/2011/02/15-rdb2rdf-minutes.html
People with action items: sdas2 souri

[End of scribe.perl diagnostic output]