Re: Follow up on today's conference call...


In our presentation

    A SQL-based Approach for Mapping Relational data to RDF

Date: 2009-12-01

Presenter: Souri Das and Seema Sundara (Oracle)


    * presentation

we had identified the requirements for mapping and then in the last four 
slides titled:
- No DB: A Simple Syntax for Classes
- No DB: A Simple Syntax for Properties
- No DB: A Simple Syntax for Constraints
- Summary
we showed how a simple syntax can be sufficient to specify the mapping 
- "SQL view-definition strings" and the corresponding RDF class names
- Columns in the view definition (referred via its position in the 
sequence of cols) and the corresponding RDF property names
  (note: foreign key (referential integrity) constraints (defined in the 
constraints syntax below) correspond to RDF properties as well)
- Constraint definitions identifying the respective column sequences for 
the keys

As we had said during the talk, since the complexity of the view 
definition gets taken care of by use of SQL language itself, the 
remaining requirements of mapping the 1) view names to RDF classes, 2) 
view columns (and referential constraints) to RDF properties, and 3) 
specifying the key definitions for the constraints (esp. for referential 
constraints), are simple and can even be expressed, for example, by 
using a simple subset of D2RQ's RDF vocabulary (d2rq:ClassMap, 
d2rq:PropertyBridge, etc., but excluding all complex vocabulary words 
such as d2rq:join, etc.)

- Souri. wrote:
> On Tue, 12 Jan 2010, Ezzat, Ahmed wrote:
>> Hello All,
>> It seems we agreed today that SQL mapping approach and customized 
>> mapping that would require no DB schema modifications need to be 
>> supported.
>> We will create two small teams working in parallel and let us call them:
>> 1.      SQL View Approach (Soren/Triplify, Ashok/Oracle, ...).  Soren 
>> and Ashok are confirmed.
>> 2.      Customized mapping approach.  None is confirmed.  Hopefully, 
>> we can get Richard/D2R, Metatomix, ...
> I can work on customized mapping approach (since I'm currently using 
> D2RQ in some of my work), but would really prefer it if 
> Richard/Howard/more could help here.
>         cheers,
>             harry
>> The goal of each team is to come up with a DRAFT requirements for 
>> their approach, thoughts on specifications, and validation to the use 
>> cases that will come out from the WG independent of these two teams.
>> This email is to solicit participation by any member into one of 
>> these teams.  Please respond before next Tuesday.
>> Regards,
>> Ahmed
>> Ahmed K. Ezzat, Ph.D.
>> HP Fellow, Business Intelligence Software Division
>> Hewlett-Packard Corporation
>> 11000 Wolf Road, Bldg 42 Upper, MS 4502, Cupertino, CA 95014-0691
>> Office:      Email:<> 
>> Off: 408-447-6380  Fax: 1408796-5427  Cell: 408-504-2603
>> Personal: Email:<> Tel: 
>> 408-253-5062  Fax:  408-253-6271

Received on Wednesday, 13 January 2010 15:36:04 UTC