See also: IRC log
<trackbot> Date: 31 May 2011
http://www.w3.org/2011/05/24-rdb2rdf-minutes.html
resolution: minutes from 2011/5/24 approved
http://www.w3.org/2001/sw/rdb2rdf/wiki/RDBNullValues
<ivan> issue-42?
<trackbot> ISSUE-42 -- How is the direct mapping suppose to handle NULL values wrt Information Preserving -- open
<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/42
<juansequeda> <Souri> My proposal: maintain data equivalence (allowing converting either way, without loss of info) => this can be done by DM 1) always generating schema triples and 2) skipping generation of triples for NULL values
<dmcneil> +1
<betehess> +1
<dmcneil> +q
enrico: acceptable but it should be proved that this is valid
ashok: what does proving mean?
<dmcneil> -q
david: for other parts of the mapping also no proofs were made
ivan: procedural problem - the main persons are not on the call
<Souri> Richard and Enrico?
<Souri> +1 to sending emails to both to confirm
<scribe> ACTION: send email to richard and enrico about finding a compromise and closing the issue [recorded in http://www.w3.org/2011/05/31-rdb2rdf-minutes.html#action01]
<trackbot> Sorry, couldn't find user - send
david: should correctness be proved?
Eric: proofs are not required
intuitive query in sparql should give the same results as in SQL
<betehess> guys, please read this email I sent a while ago http://www.w3.org/mid/1287693123.21135.419.camel@simplet
Alexandre: proof is
difficult
... proof is a lot of work but could be done
... reverse mapping is the only interesting thing
... we should be confident that the reverse way is
possible
... implementations will provide a proof-of-concept
Eric: proof do not say much about the usability
<betehess> [[
<betehess> Theorem:
<betehess> ∀ rdb-dataset ∈ RDB, ∀ sparql-query ∈ SPARQL,
<betehess> (SPARQL2SQL(sparql-query))(rdb-dataset) =
<betehess> sparql-query(RDB2RDF(rdb-dataset))
<betehess> ]]
souri: information should not be
lost
... formal proof is nice-to-have but no requirement
<betehess> Or in plain English:
<betehess> [[
<betehess> Given a SPARQL-to-SQL mapping to access an RDB dataset, the semantics of
<betehess> the translated SPARQL query executed against this particular RDB dataset
<betehess> should be equivalent to the same SPARQL query executed against the same
<betehess> RDB dataset seen through the RDB2RDF mapping.
<betehess> ]]
ashok: reversability is the proof
<juansequeda> T(Q(I)) = Q*(DirectMapping(I, S))
juan: paper about the proof is in the works for ISWC
<juansequeda> F(DirectMapping(I, S)) = I
juan: information preserving is important and proved in the paper
<ivan> issue-22?
<trackbot> ISSUE-22 -- Support for database vendor specific SQL statements. -- open
<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/22
richard sent a proposal by email
david: SQL queries should be
SQL2008 compliant
... for validation purposes projecttions should be kept track
of
... SQL queries can be created as a resource
... in addition to the text of the query it can be annotated to
be vendor specific
<dmcneil> +q
<betehess> can't we just allow people to use another RDF type than xsd:string?
eric: SQL2008 is not so
important, but the provenance and compliance of the query
should be preserved
... there should be a property to indicate the compliance of
the query (default SQL2008)
<dmcneil> -q
eric: value of the property
should be URIs
... not literals/strings
Souri: introducing URIs opens up questions about authority
Eric: creation of URIs should not be an obstacle
<dmcneil> betehess - what data type do you have in mind?
Souri: creation of URIs might be difficult due to the vast number of product variations
Ashok: group seems to agree that a property is required to annotate the queries
<betehess> dmcneil, assuming it's an URI, I was wondering if writing "SELECT ..."^^sql-vendor-uri makes sense, but I don't know...
<dmcneil> i see
Ashok: questions: 1) is it a string or a URI, 2) what values should be used
<dmcneil> +q
souri, there should be more complex data structure representing the specific product
souri: are we standardizing the product types - probably not - should be done by the users
david: sql2008 queries can be used and optimized, other queries are opaque
ashok: issue should again be discussed in the next week and resolved later
<scribe> ACTION: eric to write proposal for issue-22 [recorded in http://www.w3.org/2011/05/31-rdb2rdf-minutes.html#action02]
<trackbot> Created ACTION-133 - Write proposal for issue-22 [on Eric Prud'hommeaux - due 2011-06-07].
<dmcneil> -q
issue-18?
<trackbot> Sorry... I don't know anything about this channel
action-126?
<trackbot> Sorry... I don't know anything about this channel
http://www.w3.org/2001/sw/rdb2rdf/track/issues/18
souri: make sure that all class
and property names are meaningful
... document is modified
... renaming should be discussed
<ericP> +1 to reviewing the name change now
<Souri> class names *MapClass ==> *Map
<ericP> +1
+1
<Souri> property names *Map ==> use*Map
<Souri> example: rr:subjectMap ==> rr:useSubjectMap
<dmcneil> personally, I preferred the old naming scheme (sorry :( )
<ericP> i have no issue with _:someTriplesMap rr:subjectMap [ a rr:SubjectMap ... ]
<betehess> +1
Eric, mixed approach or "with" instead of "use"
<Souri> ericP: rr:useSubjectMap ==> rr:withSubjectMap
ivan: if capitalization is the only difference makes it quite easy to make mistakes
<Seema> prefer different names for class and property
david: property names should be kept brief, since used more frequently
ivan: when R2RML file is written, type declarations can be ommited
souri: "use" is only three letters
eric: what's the most intuitive for the user to write?
<Souri> use => verb: seems nice for predicate
Eric: intuitive predicate names are more important than intuitive class names
souri: happy to get rid of
"class" suffix
... class names are important from a pedagogic point of
view
ashok: wasn't there already agreement on issue-18?
<dmcneil> will there be an opportunity for an informal working group get-together at semtech next week?
issue-25?
<trackbot> Sorry... I don't know anything about this channel
<ericP> dmcneil, i'll be at semtech
<ericP> (and anxious to meet up)
<ericP> (sorry i didn't geek with your crew at enterprise data world)
ashok: why is issue-25 still open?
<betehess> issue-25?
<trackbot> Sorry... I don't know anything about this channel
souri: still have an action on that
<Seema> me too
<juansequeda> +1 to shout together
ashok: telco will still take place next week
This is scribe.perl Revision: 1.136 of Date: 2011/05/12 12:01:43 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/david/enrico/ Succeeded: s/alexander/Eric/ No ScribeNick specified. Guessing ScribeNick: soeren Inferring Scribes: soeren Present: Ashok_Malhotra Alexandre dmcneil Ivan Souri Seema soeren EricP juansequeda Regrets: Percy Marcelo Enrico Michael WARNING: No meeting chair found! You should specify the meeting chair like this: <dbooth> Chair: dbooth Found Date: 31 May 2011 Guessing minutes URL: http://www.w3.org/2011/05/31-rdb2rdf-minutes.html People with action items: email eric send[End of scribe.perl diagnostic output]