Hello Miroslav,
The fundamental problem is that the rule language is not expressive
enough for your use-case. The question is what is the best
solution/workaround.
The process you describe below (possibly as a result of speaking to
Ivan) involves using construct queries as 'part of' the reasoning
process. It is not completely clear to me how you intend to do this, but
I can imagine several ways, e.g.
One method would be to use an intermediate repository, where new
statements are 'prepared' by adding them, executing a construct query to
make more statements and adding these. Then copy the result in to the
main repository. (If these go in to their own context then it is easier
to delete them if necessary).
Modifying the reasoner with one or more maths functions will be a step
in the right direction, but a proper solution would involve something
more comprehensive, e.g. SPARQL.
With my current level of understanding of the issue, this is all the
advice I can give.
Regards,
barry
--
Barry Bishop
OWLIM Product Manager
Ontotext AD
Tel: +43 650 2000 237
email: barry.bis...@ontotext.com
www.ontotext.com
On 22/11/11 09:36, Miroslav Líška wrote:
Dear sirs,
let me please reopen the talk about reasoning in OWLIM/SESAME. Ivan
Peikov explained me, that custom rules at the moment do not support
SPARQL 1.1 or some math built-in functions, and he suggest me to use
SPARQL (construct).
I am afraid, that this method will be unusable. Since we want to work
with cca. 1Bilion triples that use serious inference (owl:sameAs, and
many custom rules...) the usability of SPARQL kind reasoning would
look like this:
1. I will load thousands of triples in each batch to the repository,
where the Trree reasoner will use custom rules for business logic.
2. Next I would to use SPARQL insert query above all implicit and
explicit triples, where the conditions will be rules written inside
the SPARQL 1.1
3. I would need to rereason td database again, because I created new
triples based on step 2. Moreover, the lack is that the triples
created in step two are explicit, rather than implicit.
I hope that I explained it well.
Thus let me please ask, is there possibility to write custom function
for the Trree reasoner likewise it is possible to SPARQL query?
Because with custom function in custom query I would need to use step
1, ie. simple triples loading and automated inference.
With kindest personal regards
miroslav
_______________________________________________
OWLIM-discussion mailing list
OWLIM-discussion@ontotext.com
http://ontotext.com/mailman/listinfo/owlim-discussion
_______________________________________________
OWLIM-discussion mailing list
OWLIM-discussion@ontotext.com
http://ontotext.com/mailman/listinfo/owlim-discussion