Hi,
There's no particular special support in Fuseki for working with Pellet.
To get support on Pellet logging issues and optimizing reasoning you
probably need to contact the Pellet folks.
Dave
On 13/09/12 15:46, Myunghwa Hwang wrote:
Hi, list!
I am trying to make Jena-fuseki produce inferred statements when an
ontology is uploaded to it.
The vocabulary I am using includes terms for describing some associations
between two spatial objects
(e.g., spatial pair, hasContiguityType, hasDistance, hasSpatialWeight,
etc.) and a collection of individuals instantiating those terms (e.g., a is
a spatial object; a and b are a spatial pair; this spatial pair has a
distance of 2.0, etc.).
In this vocabulary, I also defined several SWRL rules like this:
SpatialPair(?n) ^ hasContiguityType(?n,?t) ^ Rook(?t) ^
hasContiguityOrder(?n, ?o) ^ equalTo(?o, 1) -> ValidSpatialNeighbor(?n)
...
When I entered to Fuseki an ontology with a small number of individuals
(e.g., about 210 individuals), Fuseki produced all expected inferred
statements. But, when I increased the number of individuals to 4851, Fuseki
got killed due to the out of memory error. And, this error occurs when
pellet does rule-based inference. The heap size for Java was set to 3GB.
And, my configuration for Fuseki can be found here:
https://dl.dropbox.com/u/21798649/config.ttl.
A test ontology is also here: https://dl.dropbox.com/u/21798649/all.owl
Here are my questions:
1) Are there better ways to set up Fuseki so that Pellet-based inference
goes smoothly? For example, indexing or use of OntModel? If so, where can I
find corresponding documentation?
2) How can I make Pellet output its logs into the log file for Fuseki?
Any other comments or advice would be also helpful.
Thanks in advance!
-- Myung-Hwa Hwang
