On 21/04/14 22:24, Lebling, David (US SSA) wrote:
Andy and Dave,
Okay, that helped a lot. There is a place in my model creation code that invoked
OntModel.createOntology with an empty string for the URI. I changed it to be the same URI
as the Individual the model represents. (In my abbreviated sample RDF that Individual is
"http://anchor/plan#Graph1";. Making this change fixed the problem but it means
that the Individual is both a Plan and Ontology now. That will be true for all
Individuals stored as graphs.
Somehow this seems odd to me, although I noted some commentary on the web where
it was stated that many objects are both some domain-specific class and OWL
class Ontology. Is there a more correct way to do this, or is this the standard
method?
It's somewhat unusual to have the same thing be both an Ontology and an
Individual in the domain, though not technically illegal.
You may want to invent a URI for the ontology that is related to, but
different from, your other URIs. Designing the URI architecture is
often a non-trivial part of designing a linked data application.
Dave
-----Original Message-----
From: Andy Seaborne [mailto:[email protected]]
Sent: Monday, April 21, 2014 4:21 PM
To: [email protected]
Subject: Re: Fuseki/TDB: BadURIException in DatasetAccessor.putModel
On 21/04/14 19:39, Lebling, David (US SSA) wrote:
However, in this case I'm not the one who is converting a model to a payload;
it's something underneath Fuseki's DatasetAccessor.
The data going in to putModel() has a bad URI - the "" URI.
> This:
> [, http://www.w3.org/2002/07/owl#imports, http://anchor/OntS] > looks like the
subject is the URI string ""
Bad data.
> When I write to a text file, I supply a base URI.
Supplying the baseURI on writing only affects the abbreviation of URIs in the
data. They must be absolute in the data in the first place and as the example
shows, it's not.
Andy
