[redland-dev] combining models/inference support

[Dave Beckett]

On Tue, 2005-07-26 at 11:01 -0400, Vincent Carey 525-2265 wrote:
> I have just started creating an interface from R (www.r-project.org) to
> librdf.
> 
> 1) I do not see a high-level function for combining RDF models, analogous
> to the Jena add method.  Am I missing something, or must one write
> a combination function explicitly?  If the latter, is it necessary
> to check for the existence of the statement to be added in the model
> to which it will be added?

There is no union model function or class, although there are
un-implemented methods predicting it's arrival sometime
(librdf_model_add_submodel - don't try it, it won't work)
One idea would be to add a new implementation of the librdf_model class,
which did union models.  The code's already there, but the only
implementation right now is one that stores models in an librdf_storage
via the librdf_model_storage implementation.  If a librdf_model_union
implementation class was made, you'd need to implement the librdf_model
methods.

Redland won't let you add duplicate statements (unless you are using
contexts).

> 2) Are there any mechanisms for computing entailments of RDF models
> in redland, or plans to introduce such mechanisms? 

No immediate plans or time or energy for that from myself.

>  I would be interested
> to know of C/C++ resources that can do this.  For OWL inference it
> would seem that FaCT++ would be useful, but I am not clear on how
> FaCT++ resources can be used as a standalone library in conjunction
> with redland.  Any experiences with this combination?

I had a look at FaCT++ but it was such a struggle to build it, I never
got to seeing how it could be used. It'd need a C API around it
probably.  Mostly I don't know enough about this to give a convincing
answer.

Dave