RDB2RDF Working Group Teleconference -- 01 Mar 2011

<trackbot> Date: 01 March 2011

I can try to do my best

today

<mhausenblas> scribenick: boris

TOPIC ADMIN

PROPOSAL: Accept minutes

of last meeting

http://www.w3.org/2011/02/22-rdb2rdf-minutes.html

RESOLUTION: Accepted minutes of last meeting

michael: Mid march new publication
...: drafts

Direct Mapping

<mhausenblas> ACTION-98?

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

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

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

<mhausenblas> 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

<mhausenblas> close ACTION-99

<trackbot> ACTION-99 Change ISSUE-13 to postponed and add ref from ISSUE-11 to ISSUE-13 closed

<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

juan: action 99 close

michael: any news regarding this?

MacTed: nothing new

<mhausenblas> ISSUE 14?

<mhausenblas> ISSUE-14?

michael: ISSUE 14 ... ?

ISSUE-14?

<trackbot> ISSUE-14 -- Many-to-Many tables -- open

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/14

michael: anything related to this issue?

juan: let's hear richard opinion about this? ... many to many on direct mapping
... richard, you pointed out that d2r users were waiting for this

richard: what do we want to achieve with the DM? that the question

<ericP> this will be tricky without resolve without ashok and richard on the same call
...: useful thing of DM
... what people want to do with this default RDF
... sparql query for sure
... I would evaluate the DM by how easy is to do sparql query against it
....: as M2M is hard to do a sparql query
...: also there will be some theoretical cases ...
... these corner cases are not convincing
... any given tool for just pushing button for generating RDF triples
... not to generate an intermediate table

juan: how many people complain about this in d2r?

richard: no one has complaint about this

<ericP> Stonebraker's counter example: CREATE TABLE Marriages (p1 INT, p2 INT, FOREIGN KEY (p1 REFERENCES Person(ID), FOREIGN KEY (p2 REFERENCES Person(ID)));
...: since the current approach (get rid of the intermediate table) ... 5 years more or less
... no corner cases ... d2r has only one approach
... after we did the m2m, no more complains

noise

richard: cannot recall anything about the current approach

eric: it's nothing particular about this
...: for instance nothing to get the directionally
... main issue is the complexity

juan: I dont see the corner example
...: person and married

richard: order of the columns by someone
...: even arbitrarily
... we can say it is incorrect

eric: ?? modelling is feasible
...: there are cases when is optimal

MacTed: any change, change the mapping
...: then the mapping is different even if is a "bit" different

eric: if I made assertions to be tracked ...
...: I don't need to visit my code

MacTEd: dependes in the code

richard: not true in SQL

eric: I meant other thing

MacTed: q: create two tables in RDB
... join id equals identify columns
...: you have sth not resolvable?

is far from norm

eric: look at different table

juan: keep it simple
...: everyone is gona to complain
... then we have to go back
... the only way to "prove" this is richard experience
... we have to learn from this

michael: kind of lost
...: the group more or less agreed to do anything about m2m?
... are we able to come with a proposal?

juan: don't think so
... yet
... richard, do you want to inlclude it in the DM?

richard: yes

juan: include a binary table as a relationshiop
... we don't keep it simple, keep it real

<dmcneil> +q

michael: let's keep mid march ...
...: either put some work to resolve it or came to a proposal

dmcneil: only binary table?

juan: two foering keys, compound columns ...binary relation

dmcneil: ok, thanks

juan: the proposal can be we need more feedback

<ericP> s/eric?:/dmcneil:/

juan?

thanks juan!

michael: would you please juan type the proposal
...: we should try to get feedback on this ...

<juansequeda> PROPOSAL: Many-to-Many table in direct mapping is still an open issue. Publish the Direct Mapping WD with this open issue and get feedback

richard: ...extra work to write those queries ... proposes to do an extra work on the editors

eric: DM foundation, is not the final product
...: what do we have to do in R2RML
... nice mapping from SPARQL to SQL queries, how hard we make their lifes

michael: we have a concrete proposal
...: we need more input

<MacTed> +1 Proposal as written
...: as an open issue

<ericP> +1
...: any objections of the current proposal

richard: another option is to do both, put the table in the graph, and also the direct table in the graph

michael: good to include this in issue 14

eric: only subject?

michael: anywhere

<juansequeda> ISSUE-14?

<trackbot> ISSUE-14 -- Many-to-Many tables -- open

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/14

<mhausenblas> Michael: if you mentioned ISSUE-14 in a mail tracker will pick it up

michael: no objections to this proposal

RESOLUTION: ROPOSAL: Many-to-Many table in direct mapping is still an open issue. Publish the Direct Mapping WD with this open issue and get feedback

sorry

SOLUTION: ROPOSAL accepted
... PROPOSAL accepted

RESOLUTION: Many-to-Many table in direct mapping is still an open issue. Publish the Direct Mapping WD with this open issue and get feedback

accepted

<mhausenblas> ISSUE-9?

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

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

michael: issue 9, blank node... status?
... is there an action about that, yes ...
... any progress?
... either we have sth there, or we can handle next week
... it depends if we deliver sth there
... please do not deliver an hour before the call

<mhausenblas> ISSUE-10?

<trackbot> ISSUE-10 -- Hash vs Slash -- open

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/10

michael: issue 10
... status?
... juan?

eric: several emails about it
... hash with no fragment identifier
... in html are identical

juan: no preferences about this

<ivan> +1 cygri

michael: something that a user wants to provide as a configuration

<ivan> #_

richard: on the DM the closest approximation by default

<juansequeda> base_uri/table = URI for a Table

<juansequeda> base_uri/table/attribute = URI for a column

richard: argument not having _ ... because URIs looks nicer
... not everyone wants to resolve these urls

<juansequeda> base_uri/table/attribute=value#_ vs base_uri/table/attribute=value URI for a tuple

richard: this is just for resolve on the web

michael: there is an option to write R2RML to modify it

ivan: two separate issues, use # or use #_
...: already pointed in the emails
... if we don't use the "#" put an extra load on the server side

richard: assumes that the identities are non-information resources, ...

ivan: they identify in the table
... they identify a row in the table

richard: this is information resource for me

ivan: you said that you can use R2RML to modify it
... adding or removing

michael: LD deployment is important and we have to do it ...

<ericP> what's the md5sum of <People/ID=7#_> ?

michael: I don't see any strong reason for not doing that

<ericP> or the md5sum of <People/ID=7> ?

ivan: agree, if we use a #, this works in most cases ... people may not like the # because is ugly ....

richard: sorry Ivan, the row in a table is an information resource

eric: ..... trying to separate IR to real

richard: you are confusing resource and representation
... eric=you
...: we have spent some much time on this
... no new things about this ..
... many cases it doesn't matter at all ... people only has to spaqrl

ivan: way more specific that the discussion about http 14
...: the URI that identify that row is this
... agree with Richard

eric: things to look at it ...
...: is sth in the DB that is an information resource

scribe: according to richard, yes

MacTed: it is always an information resource
...: description is a document

<cygri> quoting AWWW: The distinguishing characteristic of these resources is that all of their essential characteristics can be conveyed in a message. We identify this set as “information resources.”
...: information in the database is information resource

richard: definition of information resource
... IR: The distinguishing characteristic of these resources is that all of their essential characteristics can be conveyed in a message. We identify this set as “information resources.
...: what is the case in which we can have a problem?
... access via SPARQL, no problem
... access via deferenable URI

no problem
...: only if you are on these cases,
....: not SPARQL, no dump
...: not use R2RML
... only then you will have the case

eric: compatible with LD principles
... whether or not use SPARQL, we make sure to make resolvable
... again, db records are not information resources, 303, ... there is a community that think this is an issue

richard: who?

eric: . ...

<cygri> Jonathan Reese (?)

eric: bunch of people from standford
... Jonathan Rees
... for example

<MacTed> http://sharedname.org/

michael: we won't be able to come up with sth, how important is this?

<cygri> PROPOSAL: row identifiers should be base_uri/table/attribute=value without # or #_ at the end

<mhausenblas> http://www.w3.org/TR/2010/WD-rdb2rdf-ucr-20100608/#intro-whyrdb2rdf

michael: let's see the UC document
... based on that, we need to explain in the DM ....
... I don't see why to emphasize one of them,

eric: do we have to eliminated ...

MacTed: we are talking about the DM, not the optimal mapping
... it is basic, it is the simplest
... and it will work

eric: it works both ways

MacTed: it doesn't work in all cases

OPOSAL: row identifiers should be base_uri/table/attribute=value without # or #_ at the end

<ericP> -1

<MacTed> +1 proposal as written

PROPOSAL: row identifiers should be base_uri/table/attribute=value without # or #_ at the end

<cygri> +1

<mhausenblas> +1

<dmcneil> +1

<juansequeda> +1

<juansequeda> +1 from marcelo too

<ivan> 0

eric: when people say these are not information resources ....
...: if they are, there are requirements for LD ....
... hash space reduces the number of round trips

eric: there are people

michael: I don't see any strong reason

juan: we still on track of mid march deadline

michael: no very good reason
...: if they are not happy ...

richard: regardless of the nature of the resources, the suggestion to do without the hash works
...: because you can use R2RML to modify it
... even with the argument of they are not information resources

eric: three arguments, which one are you refering to?

richard: only one
... I hear only one
... even if they are not information resources, why to do this in the DM

eric: if they are information resources, I predict we will have much problem ... if we don't treat them as information resources
...: if they aren't not information resources, we have some impact on the server, 303

MacTed: but we have R2RML

richard: 303 are compatible with LD

MacTed: your objection is gone the

eric: nonIF DM produces a graph that are both compatible to LD and efficient to access

<cygri> +1 MacTed

MacTed: default is not always efficient

+1 MacTed

eric: ok

MacTed: so remove this objection?

eric: if we say that de direct graph consithers 303 ...

MacTed: you can include it in your payload

eric: I can live with it, if we can do the rewrite

<mhausenblas> PROPOSAL: row identifiers should be base_uri/table/attribute=value without # or #_ at the end

PROPOSAL: row identifiers should be base_uri/table/attribute=value without # or #_ at the end

<mhausenblas> Michael: any objections?

RESOLUTION: row identifiers should be base_uri/table/attribute=value without # or #_ at the end

michael: so, we can close the issue right?
... DM editors pls make sure to implement this on DM and close issue 10
... do you need an action for that?
... implemented the resolution
... to implement the resolution

juan: I'll check it and do some clean as well
... hierarchical tables?

<mhausenblas> ACTION: Juan to implement decision re ISSUE-10 (remove hash) and close ISSUE-10 [recorded in http://www.w3.org/2011/03/01-rdb2rdf-minutes.html#action01]

<trackbot> Created ACTION-104 - Implement decision re ISSUE-10 (remove hash) and close ISSUE-10 [on Juan Sequeda - due 2011-03-08].

michael: DM issue: 10 (closed) 14 (postpone), 15 (open)
...:

michael: next time focus on R2RML
... action 10 is closed

michael I don't remember how to clean the minutes?

<mhausenblas> trackbot, end telecon

- DRAFT -

RDB2RDF Working Group Teleconference

01 Mar 2011

Attendees

Contents

Direct Mapping

Summary of Action Items

Scribe.perl diagnostic output