Reasoner and data graphs

[Claude Warren <cl...@xenei.com>]

I have gotten my first chance to look at building a rule base reasoner and
was wondering why updates to the data graph are not processed by the
reasoner.  My code:

@Test
public void testSubClassOf()
{
Model model = ModelFactory.createDefaultModel();

// create a resource (empty resource)
Resource config = model.createResource();

// set engine mode
config.addProperty(ReasonerVocabulary.PROPruleMode, "hybrid");

config.addProperty(ReasonerVocabulary.PROPruleSet, "etc/rdfs-fb.rules");
 Reasoner reasoner = GenericRuleReasonerFactory
.theInstance().create(config);

Graph graph = GraphFactory.createGraphMem();
InfGraph infGraph = reasoner.bind( graph );
 Node n1 = Node.createURI( "http://example.com/p1");
Node n2 = Node.createURI( "http://example.com/p2");
Node n3 = Node.createURI( "http://example.com/p3");
graph.add( new Triple( n1, RDFS.subClassOf.asNode(), n2 ));
 Assert.assertTrue( infGraph.contains( n1, RDFS.subClassOf.asNode(), n2 ));
 System.out.println( "Adding triple ");
graph.add( new Triple( n2, RDFS.subClassOf.asNode(), n3));
 Assert.assertTrue( infGraph.contains( n1, RDFS.subClassOf.asNode(), n2 ));
Assert.assertTrue( infGraph.contains( n2, RDFS.subClassOf.asNode(), n3 ));
Assert.assertTrue( infGraph.contains( n1, RDFS.subClassOf.asNode(), n3 ));

}

fails on the last Assert.  However if I change "graph.add( new Triple( n2,
RDFS.subClassOf.asNode(), n3));"
to "infGraph.add( new Triple( n2, RDFS.subClassOf.asNode(), n3));" it works
as expected.

Shouldn't the infGraph listen to the changes in the underlying graph and
update the reasoned triples accordingly or am I missing something ?

Thx,
Claude
-- 
I like: Like Like - The likeliest place on the web<http://like-like.xenei.com>
Identity: https://www.identify.nu/user.php?claude@xenei.com
LinkedIn: http://www.linkedin.com/in/claudewarren

Re: Reasoner and data graphs

[Dave Reynolds <da...@gmail.com>]

Hi,

On 15/08/12 15:39, Claude Warren wrote:
> I have gotten my first chance to look at building a rule base reasoner and
> was wondering why updates to the data graph are not processed by the
> reasoner.  My code:
>
> @Test
> public void testSubClassOf()
> {
> Model model = ModelFactory.createDefaultModel();
>
> // create a resource (empty resource)
> Resource config = model.createResource();
>
> // set engine mode
> config.addProperty(ReasonerVocabulary.PROPruleMode, "hybrid");
>
> config.addProperty(ReasonerVocabulary.PROPruleSet, "etc/rdfs-fb.rules");
>   Reasoner reasoner = GenericRuleReasonerFactory
> .theInstance().create(config);
>
> Graph graph = GraphFactory.createGraphMem();
> InfGraph infGraph = reasoner.bind( graph );
>   Node n1 = Node.createURI( "http://example.com/p1");
> Node n2 = Node.createURI( "http://example.com/p2");
> Node n3 = Node.createURI( "http://example.com/p3");
> graph.add( new Triple( n1, RDFS.subClassOf.asNode(), n2 ));
>   Assert.assertTrue( infGraph.contains( n1, RDFS.subClassOf.asNode(), n2 ));
>   System.out.println( "Adding triple ");
> graph.add( new Triple( n2, RDFS.subClassOf.asNode(), n3));
>   Assert.assertTrue( infGraph.contains( n1, RDFS.subClassOf.asNode(), n2 ));
> Assert.assertTrue( infGraph.contains( n2, RDFS.subClassOf.asNode(), n3 ));
> Assert.assertTrue( infGraph.contains( n1, RDFS.subClassOf.asNode(), n3 ));
>
> }
>
> fails on the last Assert.  However if I change "graph.add( new Triple( n2,
> RDFS.subClassOf.asNode(), n3));"
> to "infGraph.add( new Triple( n2, RDFS.subClassOf.asNode(), n3));" it works
> as expected.

Correct, that's the intended usage.

The InfGraph is a layer over the top of some other graph (or stack of 
other graphs). If you add data into the top of the layer the reasoner 
sees it immediately and can process it. If you "sneak behind its back" 
and add data at a lower layer then call infgraph.rebind() to let the 
reasoner know.

> Shouldn't the infGraph listen to the changes in the underlying graph and
> update the reasoned triples accordingly or am I missing something ?

That's not the current design. Partly that's historic and partly a 
reluctance to rely on always having complete graph listeners in order to 
use inference.

Dave