W3C

- DRAFT -

RDB2RDF Working Group Teleconference

08 Feb 2011

Agenda

See also: IRC log

Attendees

Present
+1.314.394.aaaa, mhausenblas, dmcneil, soeren, Ivan, +1.781.273.aabb, EricP, privera, +49.133.6.aacc, MacTed, boris, +1.603.897.aadd, Souri
Regrets
Ashok, Nuno, Seema, Alexandre, Richard
Chair
Michael
Scribe
ericP, mhausenblas

Contents


<trackbot> Date: 08 February 2011

<juansequeda> mhausenblas, I might not be able to call in. So I'll be just *listening* through irc

<mhausenblas> scribenick: ericP

Admin

<mhausenblas> PROPOSAL: Accept the minutes of last meeting, see http://www.w3.org/2011/02/01-rdb2rdf-minutes.html

<boris> +1

+1

<Souri> aadd is me

RESOLUTION: http://www.w3.org/2011/02/01-rdb2rdf-minutes.html is a record of the last meeting

mhausenblas: SemTech session not accepted

<mhausenblas> http://www.w3.org/2011/03/w3c-track.html

mhausenblas: can we exploit the W3C track at WWW2011?

ivan: W3C has to pick two major topics for the track. those have already been announced
... you can ask mcf, but at WWW2010, both days were consumed by the two announced topics
... i plan to be at WWW2011, happy to jump in

<mhausenblas> ACTION: mhausenb to ask Marie-Claire Forgue if there is a chance to present or discuss RDB2RDF at the W3C track at WWW2011 [recorded in http://www.w3.org/2011/02/08-rdb2rdf-minutes.html#action01]

<trackbot> Created ACTION-100 - Ask Marie-Claire Forgue if there is a chance to present or discuss RDB2RDF at the W3C track at WWW2011 [on Michael Hausenblas - due 2011-02-15].

ivan: perhaps a workshop has a slot for an e.g. panel

mhausenblas: would be happy to have an RDB2RDF sandwich board

<scribe> ACTION: ivan, mhausenblas to investigate alternatives at SemTech [recorded in http://www.w3.org/2011/02/08-rdb2rdf-minutes.html#action02]

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

R2RML status

<scribe> ACTION: ivan to investigate alternatives at SemTech [recorded in http://www.w3.org/2011/02/08-rdb2rdf-minutes.html#action03]

<trackbot> Sorry, amibiguous username (more than one match) - ivan

<trackbot> Try using a different identifier, such as family name or username (eg. ivan, imikhail)

<scribe> ACTION: herman to investigate alternatives at SemTech [recorded in http://www.w3.org/2011/02/08-rdb2rdf-minutes.html#action04]

<trackbot> Created ACTION-101 - Investigate alternatives at SemTech [on Ivan Herman - due 2011-02-15].

<mhausenblas> 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

<mhausenblas> 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

Souri: some progress

<Souri> http://www.w3.org/2001/sw/rdb2rdf/wiki/Use_of_Direct_Mapping_in_R2RML

-> http://www.w3.org/2001/sw/rdb2rdf/wiki/Use_of_Direct_Mapping_in_R2RML Use of Direct Mapping in R2RML

Souri: say we have triples map with a subject map and one or more predicate object maps
... it becomes tedious to list PO maps for each column

<mhausenblas> scribenick: mhausenblas

Michael: fine, let's give Souri a bit more time to flesh out details

Ivan: I think Michael and Souri are talking about the same thing (continuum from DM to full R2RML)

<ericP> soeren: would like to see a directive to enable default mapping

<scribe> scribenick: ericP

soeren: would like to see a directive to enable default mapping

Souri: looking at bullet one, PO maps include direct mapping
... second bullet includes r2rml:exclude
... third bullet has r2rml:includes -- start empty and include explicitly include columns

<mhausenblas> 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

mhausenblas: action remains open

<mhausenblas> 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: in the last meeting, we discussed blank nodes in the direct mapping

<mhausenblas> ISSUE-9?

<trackbot> ISSUE-9 -- Generate Blank Nodes for duplicate tuples -- open

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/9

Souri: stimulated email discussion

<boris_> why we don't collect all the possibilities on a wiki page? (regarding blank nodes)

<soeren> http://www.w3.org/2001/sw/rdb2rdf/track/issues/9

soeren: i have the impression that we are divided on bnodes
... some don't want them for linked data reasons

ivan: i feel we should remain as we have it
... direct mapping philosophy assumes another layer of massage
... if i want to push data into an LD environment, the second stage can take care of it
... if we eliminate bnodes, it will be complicated to generate URIs for what are currently bnodes
... 'cause we need to communicate to folks (via e.g. url scheme) that this node is effectively a bnode

<boris_> yes

mhausenblas, has this anything to do with r2rml?

MacTed: guidance around bnodes is that we want to avoid them 'cause they cause trouble down the line

<Zakim> ericP, you wanted to ask how non-repeatable "persistent" identifiers can be used

<mhausenblas> scribenick: mhausenblas

ericP: in the relational model we have IDs for tupels
... using PK to identify a tuple
... if we have some data via DM and we advertise some IDs other than the PK, it might no longer be valid
... so we have a unreliable URI if tuple changes

<ivan> +1 to Eric

ericP: so bNodes are the only reliable mechanism for tuples that might change
... so, I'm all for preserving the current status

<ericP> mhausenblas: our charter says we have to be linked-data-friendly

<scribe> scribenick: ericP

MacTed: there is a temporality to RDB-contained data
... the graphs captured should reflect that temporarility

<Zakim> mhausenblas, you wanted to remind on our charter

ericP: but these syntehtic identifiers can't be used

MacTed: you can have a document which describes this identifer, giving a time windows

ivan: blank nodes issue came up at RDFNext

<mhausenblas> ACTION: Ted to sum up the possibilities for generating reliable URIs for DM (to avoid bNodes) on the Wiki [recorded in http://www.w3.org/2011/02/08-rdb2rdf-minutes.html#action05]

<trackbot> Created ACTION-102 - Sum up the possibilities for generating reliable URIs for DM (to avoid bNodes) on the Wiki [on Ted Thibodeau - due 2011-02-15].

<ivan> http://bnode.w3.org/UUID-based-ID

ivan: david booth proposed a URI scheme for anonymous nodes
... if a client sees that URI, they can dereference it but get something which which isn't very informative

<MacTed> that is a potentially viable suggestion...

ivan: if we want to do this, we have to precisely define this URI scheme

<MacTed> much like the canonical variable names -- foo, bar, blat, thud...

Souri: some RDBs have a persistent identifier, e.g. Oracle's rowid
... but what if the table gets moved? we can't use it as an identifer anymore
... but we need to communicate that it may suddenly disappear

<Souri> http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Feb/0035.html

Souri: i.e. that it's not so reliable
... i tried to compare the bnode and the synthetic URI cases
... if we're returning this URI based on a uniqe sequence number from a query, or returning a bnode, neither can be used again

<MacTed> that also covers the temporal nature

<Zakim> ericP, you wanted to ask what is the motivation to assign URIs which we can't re-use?

<ivan> I think that Souri's description is indeed close to David's approach

ericP: if we don't have an identifier for a row, why do we want to pretend we have an identifier

mhausenblas: need to work out in use cases

ivan: besides saying "we don't want blank nodes 'cause the linked data doesn't like them", we need more arguments
... on the wiki page [mhausenblas's request], we need use cases

<mhausenblas> Michael: some if not all of our use cases (http://www.w3.org/TR/2010/WD-rdb2rdf-ucr-20100608/) require to have linkable resources

ivan: i understand that in many cases i prefer to use an identifier

soeren: perhaps a compromise is to allow one to use a rowid if there is one

ivan: Souri came up with a schems with an R2RML with exceptions

<Souri> +1

ivan: so can't i create a mapping which defines only row (subject) identifiers, they can do it in five lines

<MacTed> default bnode? danger lies ahead...

Souri: yes, you can create a logical table using the rowid

<mhausenblas> PROPOSAL: The DM does not try to define URI-creation, this is topic of and must be done via R2RML

<Zakim> mhausenblas, you wanted to propose a solution

Souri: per ivan's point, you can create a tiny mapping which is 99% direct mapping

<mhausenblas> PROPOSAL: The DM does not try to define URI-creation when there is no PK, this is topic of and must be done via R2RML

+1

<ivan> +1

',' should be a';'

<MacTed> +1, with eric's change

<MacTed> :-)

<MacTed> in fact, it's true -- PK is also temporal...

dmcneil, why is the URI creation when there isn't a PK so differnt from when there is one

dmcneil: why is the URI creation when there isn't a PK so differnt from when there is one?

MacTed: PK can change over time

<Souri> pkey change implies a resource disappears, a new resource appears => not much we can do about it

soeren: so ericP is assuming that PKs are stable and MacTed is assuming they're not

<boris_> nothing lasts forever

mhausenblas: if you want to have reliable URIs, we need to use R2RML

MacTed: if you push URIs out of DM into R2RML, you've killed DM
... PK may be defined in the schema or interprated in the application layer
... every DB has to have a row id, but it doesn't have to expose them
... in my perfect world, there are attributes which are valid at a point in time
... there's a named graph associated with the identifer and the named graph is valid for a given time

mhausenblas: resolution requires MacTed's response to action 102

<mhausenblas> 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

<MacTed> ACTION-102?

<trackbot> Sorry... I don't know anything about this channel

<MacTed> silly scroll...

Test Cases

<mhausenblas> http://this-db-really.does-not-exist.org/

<boris_> thanks again Eric!

ADJOURNED

next meeting: next week, next chair: ashok

<mhausenblas> Michael: regrets for the next two meetings (on travel)

The DM does not try to define URI-creation when there is no PK, this is topic of and must be done via R2RML

http://this-db-really.does-not-exist.org/

<mhausenblas> trackbot, end telecon

Summary of Action Items

[NEW] ACTION: herman to investigate alternatives at SemTech [recorded in http://www.w3.org/2011/02/08-rdb2rdf-minutes.html#action04]
[NEW] ACTION: ivan to investigate alternatives at SemTech [recorded in http://www.w3.org/2011/02/08-rdb2rdf-minutes.html#action03]
[NEW] ACTION: ivan, mhausenblas to investigate alternatives at SemTech [recorded in http://www.w3.org/2011/02/08-rdb2rdf-minutes.html#action02]
[NEW] ACTION: mhausenb to ask Marie-Claire Forgue if there is a chance to present or discuss RDB2RDF at the W3C track at WWW2011 [recorded in http://www.w3.org/2011/02/08-rdb2rdf-minutes.html#action01]
[NEW] ACTION: Ted to sum up the possibilities for generating reliable URIs for DM (to avoid bNodes) on the Wiki [recorded in http://www.w3.org/2011/02/08-rdb2rdf-minutes.html#action05]
 
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2011/02/08 18:02:05 $

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/stimulated discussion/stimulated email discussion/
Found ScribeNick: ericP
Found ScribeNick: mhausenblas
Found ScribeNick: ericP
Found ScribeNick: mhausenblas
Found ScribeNick: ericP
Inferring Scribes: ericP, mhausenblas
Scribes: ericP, mhausenblas
ScribeNicks: ericP, mhausenblas
Default Present: +1.314.394.aaaa, mhausenblas, dmcneil, soeren, Ivan, +1.781.273.aabb, EricP, privera, +49.133.6.aacc, MacTed, boris, +1.603.897.aadd, Souri
Present: +1.314.394.aaaa mhausenblas dmcneil soeren Ivan +1.781.273.aabb EricP privera +49.133.6.aacc MacTed boris +1.603.897.aadd Souri
Regrets: Ashok Nuno Seema Alexandre Richard
Agenda: http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Feb/0039.html
Found Date: 08 Feb 2011
Guessing minutes URL: http://www.w3.org/2011/02/08-rdb2rdf-minutes.html
People with action items: herman ivan mhausenb mhausenblas ted

[End of scribe.perl diagnostic output]