For largely static data then the best way to mix inference and
persistent storage in Jena is to load your data into memory, do the
inference and store the results to the TDB store. Then when you want to
use it open the TDB store as a plain OntModel (or Model) with no inference.
A brute force version of this would be something like:
OntModel om =
ModelFactory.createOntologyModel(OntModelSpec.OWL_MEM_MICRO_RULE_INF);
FileManager.get().readModel( om, "myfile");
Dataset dataset = TDBFactory.createDataset("mydir") ;
dataset.begin( ReadWrite.WRITE );
try {
dataset.getDefaultModel().add( om );
} finally {
dataset.end();
}
Then when a future program wants to access the data:
OntModel om = ModelFactory.createOntologyModel(
OntModelSpec.OWL_MEM, tdbdataset.getDefaultModel() );
In some applications you only need to support a limited range of query
patterns, in which case you can replace the add(om) stage by a more
selective add of just the results you are going to be interested in.
Dave
On 17/10/13 08:24, Dibyanshu Jaiswal wrote:
Hi!
Yes you are right! Its true that the inferencing is done by the OntModel
and not the database. I tried to set the default model of the dataset by
using dataset.setDefaultModel(om), where om is a OntModel with
OntModelSpec.OWL_MEM_RULE_INF. But in this case the programme undergoes an
error.
Can you please elaborate a bit on the solution which you say about writing
the whole OntModel to a Graph? The TDB I use is initialized (create) using
an owl file read locally, which stands to be the conceptual model for our
application and i dont want to modify that file, but at the same time want
to use inferencing.
On Fri, Oct 11, 2013 at 6:22 PM, Andy Seaborne <[email protected]> wrote:
Hi there,
If you query the dataset directly, do you see any triples? You will see
different results if you query via the inference model. The inference is
not done in the database but in the inference engine associate with the
OntModel. The dtabase does not contain the deductions.
You can store the inferred results to a database by writing the whole
OntModel to a graph in the database.
Andy
On 10/10/13 13:21, Dibyanshu Jaiswal wrote:
Here is a sample code of the above stated problem.
public class ReadTDB {
public static void main(String[] args){
// open TDB dataset
String directory = "./MyDatabases/OntologyTDB"; // My TDB created
beforehand
Dataset dataset = TDBFactory.createDataset(**directory);
//Read Model from the tdb
Model tdb = dataset.getDefaultModel();
// read the OntModel from the Model
OntModel m = ModelFactory.**createOntologyModel(
OntModelSpec.OWL_MEM,
tdb );
String sparqlQueryString=null;
sparqlQueryString = "SELECT ?s WHERE { ?s <
http://www.w3.org/2000/01/rdf-**schema#subClassOf<http://www.w3.org/2000/01/rdf-schema#subClassOf>>
<
http://www.owl-ontologies.com/**unnamed.owl#ABCD<http://www.owl-ontologies.com/unnamed.owl#ABCD>>
}";
Query query = QueryFactory.create(**sparqlQueryString) ;
QueryExecution qexec = QueryExecutionFactory.create(**query, m) ;
//LINE
OF CONSIDERATION
ResultSet results = qexec.execSelect() ;
ResultSetFormatter.out(**results) ;
tdb.close();
dataset.close();
}
}
As per the above code snippet given, and the line marked as "LINE OF
CONSIDERATION", when i pass the OntModel m as the parameter the results
are
in accordence to the inference mechanisms (such as Transitive relations)
but if I change the parameter to dataset i.e.
QueryExecution qexec = QueryExecutionFactory.create(**query, dataset) ;
and hence execute the query, the results are not the same.
As per my observation the a Query made to the Dataset/TDB directly is
unable to provide inference mechanisms as provided by the OntModel, even
when the TDB creation is done as follows:
public static OntModel createTDBFromOWL(){
Dataset dataset =
TDBFactory.createDataset("./**MyDatabases/OntologyTDB") ;
Model m = dataset.getDefaultModel();
OntModel om =ModelFactory.**createOntologyModel(
OntModelSpec.OWL_MEM_RULE_INF, m );
FileManager.get().readModel( om,"./OWLs/MyOWLfile.owl");
return om;
}
Is there some way to create a Dataset object, which is Inference enabled,
similar to creation of an OntModel like:
OntModel om =ModelFactory.**createOntologyModel(
OntModelSpec.OWL_MEM_RULE_INF, m );
so that the dataset supports inferenceing mechanisms?
On Thu, Oct 10, 2013 at 3:51 PM, Andy Seaborne <[email protected]> wrote:
On 10/10/13 10:12, Dibyanshu Jaiswal wrote:
Hi!
I am new to semantic web technologies, and have started with RDF/OWL for
making web applications.
Currently i have a requirement for accessing a Ontology (OntModel with
OWLModelSpec.OWL_MEM_RULE_INF) from an OWL file.I am also able to Store
in
local TDB, all done by JENA 2.11.0. Thanks to the Nice API and Tutorial
provided for the same.
I need to fire SPARQL queries on the model to get some fruitful results.
Once the TDB is created, in order to fire search query on the same, the
results are not as expected.
As per my SPARQL qurey, when directly made to the TDB Dataset, does not
returns results (i.e. results with inference rules) accordingly. Where
as
If the same query is fired on the OntModel (loaded from the TDB, with
OWLModelSpec.OWL_MEM_RULE_INF ) itself the results are found to be as
expected.
Generally, showing the details of what you are doing makes it easier for
people to provide answers. The details matter :-)
How do I solve the problem of making queries directly to the Dataset and
not to OntModel with inference rule enabled?
Inference is a characteristic of the model (in RDF inference is within
models/graphs, not between graphs).
You need to create an ont model backed by a graph from the TDB store.
Andy
Please help!!
Thanks in Advance!!
