Hello,
I experimented with the new reasoning features and there are some issues.
I already solved them and just want to give you some feedback.
The features and the speed of reasoning are quite awesome, good work <thumbs up>.

Here is one small question:
If my:trans is a transitive Property and I query:
?s my:trans ?class. ?class a my:Person.
Will the inference engine's  answer be complete and combinable?

Here is a list of strange errors I encountered with V6.1.0.
I can provide you with the data and test queries, if you are interested.
I don't need answers for these, as for me they are already fixed.

1. CREATE BITMAP index RDF_QUAD_PSOG on DB.DBA.RDF_QUAD (P,S,O,G);
Other indexes worked, but the PSOG messed up the DB. After restarting I got the message: Error executing a server init statement : 22023: JSO_LOAD_GRAPH has not found expected metadata in the graph -- DB.DBA.RDF_QUAD_FT_UPGRADE ()
and executing a query with " define input:inference" I got this message:
JSO_LOAD_GRAPH has not found expected metadata in the graph
So server was loading data and index well, but after restarting it didn't work any more.

2. At first, I loaded an Ontology for reasoning, containing some blanknodes:

my:SyntacticNode rdfs:subClassOf _:bn1 .
_:bn1 owl:minCardinality "1"^^xsd:nonNegativeInteger .
_:bn1 owl:onProperty my:hasTag .
_:bn1 rdf:type owl:Restriction .

Whenever I was querying some high classes in the hierarchy, the query took over 400 seconds and brought the wrong result. So there might be a glitch there.

3. Inferencing doesn't seem to be compatible with Vectors, so for the pattern:
?instance a ?class
this still works:
Filter ( ?class = my:Person )
But these do not:
Filter ( ?class = my:Person || ?class = my:Student )
Filter ( ?class in ( my:Person, my:Student) )

I'm not sure if this is a bug, might be a new feature.

Dataset:
8 million triples instance data
About 1000 schema triples with subclassof, inverseof and transitive properties


Regards,
Sebastian, AKSW
--

Dipl. Inf. Sebastian Hellmann
Department of Computer Science, University of Leipzig
Homepage: http://bis.informatik.uni-leipzig.de/SebastianHellmann
Research Group: http://aksw.org


Reply via email to