W3C

- DRAFT -

RDB2RDF Working Group Teleconference

01 Dec 2009

See also: IRC log

Attendees

Present
Regrets
Chair
Michael
Scribe
soeren

Contents


 

 

<trackbot> Date: 01 December 2009

<scribe> scribe: soeren

<scribe> scribeNick: soeren

Admin

approval of minutes

<mhausenblas> http://www.w3.org/2009/11/17-RDB2RDF-minutes.html

approval of the minutes from 17th of Nov

any additions or changes?

<mhausenblas> PROPOSAL: accept minutes

<hhalpin> +1

<mhausenblas> http://www.w3.org/2009/11/24-RDB2RDF-minutes.html

<hhalpin> +1

no objections

all minutes approved

eric suggests the addition of a discussion of the datatypes issue to the agenda

any other additions to the agenda?

<Souri> The latest version (with minor changes) is at http://www.w3.org/2001/sw/rdb2rdf/wiki/images/4/43/RDB2RDF_Map_viaSQL_w3c_final.pdf

start of the presentation

presentation by oracle

<jsequeda> I apologize for slacking off on the DDL coverage http://www.w3.org/2001/sw/rdb2rdf/wiki/Requirements/DDLCoverage Thanksgiving holiday! I'll get back to that this week!

slide3 describes the architecture

architecture consists of a database and a client side

mapping specification maps relational data into classes and instances

focus is on the client side

SQL engine is not required

if available the SQL language processor generates the RDF schema and RDF instance objects

<hhalpin> Just go to slide 7

<hhalpin> for those without animation

SQL queries specify a transformation

by using foreign and unique key constraints

<Ahmed> It seems lines are mutted.

approach supports comprehensive DBs with views and constraints as well as simple structured data

<hhalpin> Line 31 is muted due to excessive noise, not sure who it is...

if a strong DB is available additional language is not required

soeren: what's meant with no-db

refers only to the client side

the no-db approach works on extracting RDF from JDBC results

an example view is mapped to an RDF class by creating an instance for each row

special columns can indicate class membership

multiple types are possible by creating several rdf:type columns

slide 13

<hhalpin> ?

<MacTed> +1 for page numbers *on* the slides going forward

<MacTed> his PDF viewer is "paper" a/k/a "hardcopy"

<MacTed> title only works when there's no content-change/redrawing

PROPOSAL: by ahmed subsequent presentations should be formated right and contain page numbers

<MacTed> page numbers are *trivial* to add when building presentations

<MacTed> and save lots of trouble

<MacTed> well, nothing's easy in Slidy... ;-)

column aliases are used for mapping to RDF properties

alter view and adding constraints allows to add unique keys

<hhalpin> wondering about "<xyz.com/emp/etype/rdf:type>"

foreign keys are used for mapping to object properties

view names correspond to class names

orri: how to to create instances with data from multiple tables

<hhalpin> So, to summarize (I think!): name of views -> URIs for class, columns in view -> instance URIs, foreign keys -> dataType properties..

orri: what about different types, how are subclass-superclass structures handled

??: is there some development environment

for now just a proposal, no tools available

the approach is very similar to Triplify

SQL should not be reinvented inside the mapping language

its possible to create several rdf:type column aliases

slide 23

different type column aliases are modelled as supproperties of rdf:type

slide 24

<Marcelo> I am also lost ...

<hhalpin> Everyone, re-download the slides if you are missing

<Marcelo> what is the title of the slide?

<MacTed>  38&&

<MacTed> $3

<hhalpin> the slides that says "SPARQL Query: Involving generic rdf:type"

<MacTed> ack. sorry.

<hhalpin> On page 24.

the subproperties of rdf:type help the SPARQL2SQL processor to find the right table to query

slide 26

view names can contain namespace prefixes

slide 27

now, no DB features available at the client side

we have to create a mapping between view names and SQL select queries

slide 30

constraint information is additionally specified

slide 31

mapping of columns to properties is done with a special data structure

slide 32

proposal for a syntax to map SQL selects to classes

slide 33

similar syntax for annotating result column

<hhalpin> Could be a chunk of code at the other end, correct? Just JDBC-compliant code...

<MacTed> started with a VIEW approach, more than with a DB approach...

soeren: both (the no-DB and DB) approach could be easily merged, into one, where the DB version would be just simple SELECT * queries

??: approach appears to be similar to property tables

relational data is not modeled as a triple view

approach is very similar to triplify

less similar to ultrawrap

D2RQ does more i.e. reinventing SQL

we should employ SQL to do relational transformations

inverse properties are important

range information can be derived

from the result column datatypes

slide 35

summary: SQL query is opaque

either views or SQL is mapped to instances

<ericP> jsequeda, care to linger to test something for me?

<jsequeda> ericP: yeah, no problem

ahmed: what's the status of the work?

prototype is not yet ready for the public, but working

orri: did you do measurements?

planned, but will take more time

ahmed: how is efficient query optimization performed

<ericP> 1-

optimization is performed by the SQL optimizer

all the work is pushed down to the DB

indexes can be used as hints

dan: rumour at ISWC - Oracle has a product in this space

oracle supports RDF data already since 3 releases

AOB

<mhausenblas> next week is http://www.w3.org/2001/sw/rdb2rdf/wiki/Initial_Round_of_Presentations#SWObjects

<mhausenblas> ericP?

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2009/12/01 18:07:55 $

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/classes/tables/
Found Scribe: soeren
Inferring ScribeNick: soeren
Found ScribeNick: soeren

WARNING: No "Present: ... " found!
Possibly Present: Ahmed Ashok BenSzekely IPcaller Kingsley_Idehen LeeF MacTed Marcelo OpenLink_Software P15 P30 P31 P32 P34 PROPOSAL RDB2RDF Seema Souri aabb aacc aadd aaee aaff aagg aahh angela_UNITN dan ericP hhalpin iv_an_ru joined jsequeda mMetatomix mhausenblas nunolopes orri scribeNick soeren summary trackbot
You can indicate people for the Present list like this:
        <dbooth> Present: dbooth jonathan mary
        <dbooth> Present+ amy

Found Date: 01 Dec 2009
Guessing minutes URL: http://www.w3.org/2009/12/01-RDB2RDF-minutes.html
People with action items: 

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.


[End of scribe.perl diagnostic output]