W3C

- DRAFT -

RDB2RDF Working Group Teleconference

25 Jan 2011

Agenda

See also: IRC log

Attendees

Present
+1.314.394.aaaa, +49.133.6.aabb, [IPcaller], mhausenblas, +1.603.897.aacc, Ivan, Ashok_Malhotra, +575737aadd, juansequeda, boris, +1.603.897.aaee, dmcneil, Seema, souri, soeren, +1.781.273.aaff, Alexandre, nunolopes, MacTed, privera, EricP, cygri, +1.562.249.aagg, Marcelo
Regrets
Chair
Michael
Scribe
boris, mhausenblas

Contents


<trackbot> Date: 25 January 2011

<boris> I can try to do it

<mhausenblas> scribenick: boris

<souri> who is on the phone?

<souri> aacc is me

<MacTed> :-)

<privera> great thanks

ADMIN

<mhausenblas> PROPOSAL: Accept the minutes of last meeting, see http://www.w3.org/2011/01/18-rdb2rdf-minutes.html

PROPOSAL: Accept the minutes of last meeting

<juansequeda> +1

<betehess> +1

http://www.w3.org/2011/01/18-rdb2rdf-minutes.html

RESOLUTION accepted

RESOLUTION: last telcon minutes accepted by the working group

<MacTed> thanks, ivan

Review of R2RML Test Cases

<mhausenblas> scribenick: mhausenblas

http://www.w3.org/2001/sw/rdb2rdf/wiki/TestCases

ACTION-91?

<trackbot> ACTION-91 -- Michael Hausenblas to move accepted TC from wiki to http://www.w3.org/2001/sw/rdb2rdf/test-cases/ -- due 2011-01-25 -- OPEN

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/91

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

<cygri> just joined. sorry i'm late

ACTION-81?

<trackbot> ACTION-81 -- Juan Sequeda to and Eric to draft Direct Mapping TC -- due 2011-01-25 -- OPEN

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/81

<boris> http://www.w3.org/2001/sw/rdb2rdf/wiki/TestCases

Boris: now under review, now
... asked Juan, Eric and Marcelo to contribute
... any questions?

<scribe> scribenick: boris

R2RML status

http://www.w3.org/2001/sw/rdb2rdf/r2rml/

souri: check two actions regarding this
... one comment from Ivan

issue number ?

michael: go to the actions ...

souri: issue 16
... ISSUE-16?

<souri> ISSUE-16?

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

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

souri: issue with the subjectmap, cardinality
... we should allow multiple subjectmaps
... one row can have multiple subjects, we allow it
... it is multiple cardinality
... comment from Ivan
... the group has to decide if this make sense
... another action?, what happen if there is no subject map?
... action related to connect r2rml direct mapping
... what do we do in this particular case?
... we have to use direct mapping? how to specify the direct mapping?
... we have some ideas to discuss here to the group
... opinion to allow multiple subjectmaps ..

cygri: it will pretty clear
... no harm
... on the other hand is not useful
... no use cases to do that
... it is trade off being flexible or to find a useful case
... it is also consider in d2r ... depending on the particular cases
... for example in equivalent classes ...

<mhausenblas> Michael: KISS principle applies - let's mark it for R2RML v2.0

cygri: no harm, but no benefit

michael: ... if we don't have a clear use case ... let's mark it for version 2

noise

souri: e.g. owl:sameAs ... multiple subjects .... "not materialized" ...
... this is one utility, but he is ok with that

ivan: what is the process should do?
... if it is only one, I can live with that, but we have to say something if there are two

souri: I like the error idea

michael, is it a erro?

<ericP> ┌────────┬────────┬──────────┬────────────┬────────────────────┐

<ericP> │ ?FNAME │ ?LNAME │ ?CONTACT │ ?PHYSICIAN │ ?PHYSICIAN_CONTACT │

<ericP> │ Bob │ Smith │ 18 │ Dr. Suess │ 23 │

<ericP> └────────┴────────┴──────────┴────────────┴────────────────────┘

michael: an action for this, what will happen in that case?

<ericP> _:_1 a foaf:Person ; foaf:givenName "Bob" ; foaf:familyName "Smith" ; :doc :_2 .

<ericP> _:_2 a :Physician ; :contact Addres:_23 .

ericP: I show the scenario for this

cygri: one thing: the sameAs scenario ... not sure if I recommend this .. but it is good point
... second: agree with Ivan to address cases like this
... error? or what happens?
... so get back to the specification and check how to it
... create an action?

michael: create wiki page, issue, action? ... just not forget it

cygri: volunteer to create a wiki page for this

<mhausenblas> ACTION: Richard to capture this (multiple subject maps) and other (future) potential error cases in the Wiki (ISSUE-16) [recorded in http://www.w3.org/2011/01/25-rdb2rdf-minutes.html#action01]

<trackbot> Created ACTION-96 - Capture this (multiple subject maps) and other (future) potential error cases in the Wiki (ISSUE-16) [on Richard Cyganiak - due 2011-02-01].

<mhausenblas> ACTION-88?

<trackbot> ACTION-88 -- Souripriya Das to adds issues to http://www.w3.org/2001/sw/rdb2rdf/track/issues/ , one for Ashok's concern and one re each of the open, inline-Issues in http://www.w3.org/2001/sw/rdb2rdf/r2rml/ -- due 2011-01-25 -- OPEN

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/88

michael: with this action 88 can be closed?

<Seema> Yes

souri: I added the issue ...

<Seema> i.e. action-88 can be closed

seema: yes, it can be closed

<mhausenblas> close ACTION-88

<trackbot> ACTION-88 Adds issues to http://www.w3.org/2001/sw/rdb2rdf/track/issues/ , one for Ashok's concern and one re each of the open, inline-Issues in http://www.w3.org/2001/sw/rdb2rdf/r2rml/ closed

michael: more actions?

<mhausenblas> ACTION-83?

<trackbot> ACTION-83 -- Ashok Malhotra to review R2RML ED -- due 2011-01-25 -- OPEN

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/83

michael: ACTION-83

ashok: I'm late

michael: ACTION-92

<mhausenblas> ACTION-92?

<trackbot> ACTION-92 -- Richard Cyganiak to address the R2RML - DM connection -- due 2011-01-25 -- OPEN

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

cygri: not yet

ACTION-93

<mhausenblas> ACTION-93?

<trackbot> ACTION-93 -- Souripriya Das to identify not-mapping vs. default-mapping issues in R2RML/DM (Wiki or via mail) -- due 2011-01-25 -- OPEN

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

souri: we discussed with Seema

<ericP> ivan, is the paste i uttered above near what you were thinking about re: multiple subjects?

souri: the idea is to be a gradual specification of the ...

<ericP> i was trying to tease out whether you want repeated properties, or generally just want to partition the properties amounst the created subjects

souri: if we look at the r2rml what are the things we can skip? ... for example subjectmap ...

<ivan> eric, not exactly. At least my idea was that several subjects gets exactly the same predicates

souri: what implies, not to specify the things in subjectmap and rely on direct map
... do what the direct map can do to get the subjectmap
... other: .... our columns to be predicateobjectmaps
... use direct map to specify the subject and predicate
... only a subset of the columns to the direct map
... exclusion list

<betehess> mhausenblas++

michael: hard to follow

<souri> rr:subjectMap rr:directMapping

souri: the idea is something like rr:subjectMap rr:directMapping
... similar to the predicateobjectmap

<souri> rr:predicateObjectMaps rr:directMapping

souri: rr:predicateObjectMaps rr:directMapping
... we can refine this and create an inclusion list
... or except, exclusion
... also can be applied to the refPredicateObjectMap
... better to write an email or in the wiki

michael: what is the overall approach
... and then we can accept

souri: three points of this subjectmap, predicatemap, refpredicatemap

cygri: ... one table for direct, two tables for custom
... we have to find out something easy to understand on all these levels

ivan: what would be really the use case for something like that?

<souri> hooking points for direct-mapping and then refine using inclusions list and exclusions list and '*'

ivan: it's relative complicated ?

souri: use case, let's say we have 10 columns in a table, I only want to customize two columns and for the rest direct mapping

ivan: I understand, but in real life if I use direct mapping the predicates generated are really not very nice, and they can be transformed by sparql
... but if I use r2rml it will be the final rdf

<ericP> i expect this will come up most frequently when folks are doing incremental updates to their r2rml mapping

<ericP> (ivan)

ivan: in practice I would not use that, but maybe I'm wrong

<soeren> +q

michael: we need to specify what do we should do?

cygri: in real life, we have database 64 tables and convert them to RDF, big job,
... pick one place to start, start with important tables and do it in a incremental way

<soeren> yes

cygri: one way to support this is direct mapping, and get the default rdf, and the "modify" it

<souri> example: all tables owned by this user modulo an exclusion list

cygri: the other approach would be: all the tables with direct mapping except employee table, in this table all the columns mapped except some cols
... it works huge database, at least in d2r
... if I have an empty r2rml, I get the direct mapping and the data from that mapping

soeren: it's important to have the incremental approach r2rml
... one option is to have shortcuts
... specify what columns to what properties and the direct mapping is used for that mappings

<cygri> lol

soeren: let's try to look to see how to simplify ... avoiding to create a lot of stuff

<cygri> well done ericP

<MacTed> /kick ericP

soeren: in this way the mapping language will be more appealing

ivan: michael?

<souri> Specified at three levels: (all tables in) DB, (all tables owned by a) user, (all columns in a) table + refined with INCLUSIONS/EXCLUSIONS/*

michael: sorry

<Zakim> ericP, you wanted to suggest that it delays inventing schema, which can move things along

ericP: motivating use case, 64 tables and we want to express by using foaf or/and dc

<cygri> +1 to ericP

michael: good motivation

ACTION-94

ACTION-94?

<trackbot> ACTION-94 -- Sören Auer to create Wiki page with options for disambiguation interface for well-known entities identifier -- due 2011-01-25 -- OPEN

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/94

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

soeren: I created the wiki page

<mhausenblas> close ACTION-94

<trackbot> ACTION-94 Create Wiki page with options for disambiguation interface for well-known entities identifier closed

soeren: and it is linked from the linked data aspects

<mhausenblas> yes, ivan, should be open

soeren: wrote 3 approaches to solve this

first one: entity resprestation on the database columns

<mhausenblas> Michael: I ask the R2RML editors to change status of ISSUE-16 to open

scribe: second mapping funcitons

third: using entity mapping table
... the easiest way is the r2rml core editors ... database column variants ..
... create uris based on namespaces prefixes predifned

souri: r2rml allows to do that

michael: thanks soeren
... how to proceed now?

soeren: we should discuss which of the variants we need?
... the first one works good
... it depends on the scenarios

cygri: third approach probably won't work with r2rml
... the main point is to store the alignment on the database, and this should work
... in scenarios where we cannot modify the database
... second approach: the main problem where is the implementation of the functions are coming from
... from the spec or free implementation
... java class that implement the function, or stored procedure or web service

michael: we don't want to go in much detail regarding the implementation
... for example an xslt

cygri: in d2rq we can specify a java class

soeren: the processor has to understand whether is a java class, web service ... etc

cygri: you can say, expected input, output and uri
... that would be possible

michael: we have to gather a bit of experience on implementation
... for example dbpedia lookup, and the possible problems we can have

soeren: second option: one problem function one to one
... second problem: could be cases we can have two syntaxes variations of a uri

ericP: rdf says two different uris two different resources

michael: usually you have to define a normalization process

ericP: I would not expect that ... on normalization

soeren: extract a part of the URI ... in this case mapping functions won't work any more

michael: don't think to worry about it, I'm with ericP

<ericP> FILTER ( tolower(scheme1) = tolower(scheme2) )

cygri: volunteer for the second one ...draft a simple interface .. and gather implementations experiences
... to include cases from d2rq
... on the wiki page

<mhausenblas> ACTION: Richard to look into D2RQ implementation and update http://www.w3.org/2001/sw/rdb2rdf/wiki/Entity_disambiguation [recorded in http://www.w3.org/2011/01/25-rdb2rdf-minutes.html#action02]

<trackbot> Created ACTION-97 - Look into D2RQ implementation and update http://www.w3.org/2001/sw/rdb2rdf/wiki/Entity_disambiguation [on Richard Cyganiak - due 2011-02-01].

<ericP> soeren, i understand the motivation, and sympathise with the desire for a smarter notion of equivalence, but i think it's unlikely ('cause it's expensive) so i wanted to demoralize you as soon as possible

cygri: the third approach should be included as well

souri: David sent some mails about sql query specification

<mhausenblas> http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Jan/0069.html

souri: constructing portions of the logical table. .. it would be a good idea

<mhausenblas> ACTION-85?

<trackbot> ACTION-85 -- Ted Thibodeau to review Direct Mapping ED -- due 2011-01-25 -- OPEN

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/85

<mhausenblas> ACTION-95?

<trackbot> ACTION-95 -- Juan Sequeda to create issues for DM draft -- due 2011-01-26 -- CLOSED

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/95

Direct Mapping status

michael: a lot of issues
... so next time to check all the issues there

<mhausenblas> http://www.w3.org/2001/sw/rdb2rdf/track/issues/raised

michael: for the next week a lot of direct mapping issues

<mhausenblas> close ACTION-95

<trackbot> ACTION-95 Create issues for DM draft closed

ashok: to ten minutes on the issues
... and classify them into simple, not simple, etc.

betehess: add new rules

marcelo: we will do it for next week

<MacTed> these including all Juan's issues? http://www.w3.org/2001/sw/rdb2rdf/track/issues/raised

juan: face to face with marcelo

michael: let's do what ashok suggested

ISSUE-9?

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

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

michael: any comments?

ashok: I haven't look at this

juan: let's do it over email

michael: have a look on all of these issues

<betehess> can I add issues?

michael: and next time we will check then

<betehess> you'll hate me

thanks

<mhausenblas> [meeting adjourned]

thanks, I forgot it ...

Ok, I'll send them

<mhausenblas> trackbot, end telecon

Summary of Action Items

[NEW] ACTION: Richard to capture this (multiple subject maps) and other (future) potential error cases in the Wiki (ISSUE-16) [recorded in http://www.w3.org/2011/01/25-rdb2rdf-minutes.html#action01]
[NEW] ACTION: Richard to look into D2RQ implementation and update http://www.w3.org/2001/sw/rdb2rdf/wiki/Entity_disambiguation [recorded in http://www.w3.org/2011/01/25-rdb2rdf-minutes.html#action02]
 
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2011/01/25 18:04:32 $

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/cygri it/cygri: it/
Succeeded: s/welll/well/
Found ScribeNick: boris
Found ScribeNick: mhausenblas
Found ScribeNick: boris
Inferring Scribes: boris, mhausenblas
Scribes: boris, mhausenblas
ScribeNicks: boris, mhausenblas
Default Present: +1.314.394.aaaa, +49.133.6.aabb, [IPcaller], mhausenblas, +1.603.897.aacc, Ivan, Ashok_Malhotra, +575737aadd, juansequeda, boris, +1.603.897.aaee, dmcneil, Seema, souri, soeren, +1.781.273.aaff, Alexandre, nunolopes, MacTed, privera, EricP, cygri, +1.562.249.aagg, Marcelo
Present: +1.314.394.aaaa +49.133.6.aabb [IPcaller] mhausenblas +1.603.897.aacc Ivan Ashok_Malhotra +575737aadd juansequeda boris +1.603.897.aaee dmcneil Seema souri soeren +1.781.273.aaff Alexandre nunolopes MacTed privera EricP cygri +1.562.249.aagg Marcelo
Agenda: http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Jan/0056.html
Found Date: 25 Jan 2011
Guessing minutes URL: http://www.w3.org/2011/01/25-rdb2rdf-minutes.html
People with action items: richard

[End of scribe.perl diagnostic output]