Hi Haibo,
indeed it is a bug in SwiftOWLIM when it restores the KB from persistence.
The cause for it is because, during repository initialization, we enumerate
the contents of the 'collections' subfolder,but we check only for files
ending with '.owlim' or '.owlim1'.
It is that way because the naming convention we choose for these files is
somehow stupid - it is based on the localname of the predicate and then we
just add a digit at the end of the '.owlim' extention until the filename
became unique. So when your dataset has predicates that share the same
localname(or part of it), we end up generating filenames with extentions
like '.owlim4' or ''owlim.5' etc. so the next time when the repository is
initialized those are ommited.
Until we prepare and publish the new release of SO a simple workaround could
be to give an unique names for the files stored in
<storage-folder>/collections so that the filenames end with '.owlim'.
But this way , altough no statements will dissapear, you will end up with
some garbage in that folder (obsolete files that we didn't remove when the
respository is shutdown next time - mostly named '.bak') ... so if you are
short of time I could build you a trree jar with that particullar fix and
send it you. So, in case you opt for that, just send me an email to do that
for you.
HTH,
Damyan Ognyanov
Ontotext AD
Dear Ivan,
We are using Sesame2.3.0 and OWLIM-3.0beta12 to create a sophisticated
RDF repository. However, we experienced difficulties in using OWLIM
persistent. The persistent loses a lot triples in our case.
Attached are the results and the simple test code and the triples we
used to demonstrate the problem.
Here is what we did:
>>>
>>> 1. Load the N-Triples. (run-1)
>>> 2. Dump the repository into n-triple and trig files.
>>> 3. Extract triples without b-node and sort it.
>>>
>>> 4.Edit the simple Java code (TestOwlim.java) not to read in the
>>> N-Triples
(cpmment out line 75,76 and 83). (run-2)
>>> 5.Run the code again. So the repository is populated from the OWLIM
>>> persistent instead read in from the external file.
>>> 6.repeat step 2 and 3.
>>>
>>> 7. diff over the files generated from step 3 and 6 and save the result
>>> to
>>> a
>>> file.
Best
--
Haibo Liu PhD.
Data Library
International Research Institute for Climate and Society
The Earth Institute at Columbia University
113D Monell, LDEO, 61 Route 9W, PO Box 1000, Palisades, NY 10964-8000
Phone: 845-680-4416 FAX: 845-680-4864
_______________________________________________
OWLIM-discussion mailing list
[email protected]
http://ontotext.com/mailman/listinfo/owlim-discussion
