Need to be able to address the subject URI as a field in the SQL table.

[Claudenw]

Need to be able to present the subject URI as a field in the table.

Select *
from People, College
where People.almaMater = College.?

alamaMater is an objectProperty with a range of College. However what is the property of College that I would use because in SPARQL terms I really want to use

Select *
{?people :almaMater ?college .
?people a :Person .
?college a :College }

[peterjohnlawrence]

To avoid creating a pseudo-column one could support a SQL inner join

Select * from People inner join College using almaMater

Of course this means that a simple join across the FK-PK would not be supported, but perhaps a SQL 'inner join' is closer to the intent of a OWL objectProperty

[Claudenw]

If you think about it, all the columns are pseudo columns. There is nothing in the driver that requires that tables be based on classes and properties, that is just the simplest implementation. It is possible to create a schema builder that is hard coded, or uses specific values of certain properties, or uses some other algorithmic process to group resources. Adding a column that contains the subject property in the current schema builder mechanism is equivalent to providing a primary key for each table. I believe that it makes the most sense and may improve performance.

[peterjohnlawrence]

Indeed, and as we go further it would be good to create an RDF-to-Entity
mapping ontology, and then those entities are published via JDBC. I have an
outline of such an ontology, but it needs work to align it with your
JDBC4SPARQL.

BTW is 'NameUtils' missing from GitHub; I have only looked briefly but it
was throwing that error. Unfortunately have not been able to spend the time
that I would like on this in the last few days, but that does not imply
lack of interest. I appreciate you working on those your fixes

Peter

On Sun, Mar 17, 2013 at 7:02 PM, Claude Warren [email protected]:

If you think about it, all the columns are pseudo columns. There is
nothing in the driver that requires that tables be based on classes and
properties, that is just the simplest implementation. It is possible to
create a schema builder that is hard coded, or uses specific values of
certain properties, or uses some other algorithmic process to group
resources. Adding a column that contains the subject property in the
current schema builder mechanism is equivalent to providing a primary key
for each table. I believe that it makes the most sense and may improve
performance.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/1#issuecomment-15033075
.

Peter J. Lawrence

---

Mobile: +1 330 631 3772 | Phone: +1 330 342 0582 | Skype:
PeterJLawrence
Email: [email protected] | Web: www.inova8.com
_LinkedIn: _http://www.linkedin.com/in/peterjohnlawrence

[Claudenw]

A good start on a mapping ontology (albeit from relational to RDF) is http://www.w3.org/2001/sw/rdb2rdf/

Please open an enhancement to pursue the ontology discussion further.

[Claudenw]

Yes, the NameUtil was missing along with 2 test files. Fixed now.