Gert,
As Greg pointed out, UIMA is just an API spec with Java (and soon C++)
Implementation(s), but you can iterate over UIMA's "CAS" annotations and
then use another existing API, Jena, to create and persist RDF as XML or
to a DB via JDBC:
http://jena.sourceforge.net/
This has a class Triple for RDF statements, an interface RDFXMLWriterI,
and so on.
Your requirement sounds like such a piec of code might be useful for
more people as well, so you might want to consider contributing your
solution to the UIMA codebase, which would have the benefit to the
community that they don't need to re-write that glue code, and to you
that you would get free extensions and bug patches from other people who
choose to use it.
Best regards,
Jochen
-----Original Message-----
From: Greg Holmberg [mailto:[EMAIL PROTECTED]
Sent: Monday, August 04, 2008 7:13 PM
To: [email protected]
Cc: Villemos, Gert
Subject: Re: AW: AW: Using UIMA for structured data sources
Gert--
UIMA does't store at all. It's just an API you call--document in,
annotations out. That is to say, Java objects. What you do with those
returned objects is your business. There's example code that can write
the annotations to an XML file (one XML file for each input document).
If you want to write the annotations to a database, a search engine, an
RDF store, etc. you'll have to write that code. UIMA knows nothing
about RDF or OWL.
Greg
-------------- Original message ----------------------
From: "Villemos, Gert" <[EMAIL PROTECTED]>
> Luckily we have included some pretty tough semantic / linguistic
> experts in the project.
>
> Another question;
> You mention that we need a UIMA-to-RDF converter. I had assumed that
> Apache UIMA stored the data graph in RDF format... as this is
> apparently not the case; which format is UIMA using?
>
> Thanks,
> Gert
