Re: [Bigdata-developers] OWL Inference

SourceForge Headquarters 1320 Columbia Street Suite 310 San Diego, CA 92101 +1 (858) 422-6466

Jürgen,

If you want to look into this, per Mike's comment, there are two implementations of "BaseClosure":

com.bigdata.rdf.rules.FullClosure [1]

and

com.bigdata.rdf.rules.FastClosure [2]

The "fast closure" algorithm is based on [3].  It's possible that some of the ordering of the OWL rules is incorrect, which might account for the problem you are seeing.  You can test this easily enough by changing to the FullClosure algorithm.  If you want to look at the code for [2] you might be able to spot the problem there.

Thanks,
Bryan

[1] http://www.bigdata.com/bigdata/docs/api/com/bigdata/rdf/rules/FullClosure.html
[2] http://www.bigdata.com/bigdata/docs/api/com/bigdata/rdf/rules/FastClosure.html
[3] "An approach to RDF(S) Query, Manipulation and Inference on Databases" by Lu, Yu, Tu, Lin, and Zhang.", http://www.cs.iastate.edu/~tukw/waim05.pdf.

> -----Original Message-----
> From: Jürgen Jakobitsch [mailto:jak...@pu...] 
> Sent: Saturday, May 01, 2010 9:06 AM
> To: Mike Personick
> Cc: big...@li...
> Subject: Re: [Bigdata-developers] OWL Inference
> 
> hi mike, thanks for your time,
> 
> i will check the source anyway - since i'd really like to use 
> bigdata, notably i'm really interested in some of the 
> features like scalability and provenance.
> 
> also let me note, that i think that bigdata source code is 
> very well done!
> 
> currently i have two options 
> 
> 1. owlim - which infers correctly, but has persistence 
> problems in some cases (so not suitable for production use) 
> amongst others 2. bigdata - which has infer problems, but has 
> features it has and which i consider very interesting
> 
> if i find something, i sure let you know.
> 
> wkr turnguard.com/turnguard
> 
> ----- Original Message -----
> From: "Mike Personick" <mi...@sy...>
> To: "Jürgen Jakobitsch" <jak...@pu...>, 
> big...@li...
> Sent: Friday, April 30, 2010 9:34:20 PM
> Subject: RE: [Bigdata-developers] OWL Inference
> 
> Jurgen,
> 
> I just wanted to let you know that while I have not gotten to 
> the bottom of the problem yet, I agree with you that there is 
> indeed a problem.
> OWLIM's answer looks more correct to me than does our answer. 
> I don't understand why we are seeing (narrower, TechnoPop) 
> and (narrower,
> ElectroPop) since narrower is not itself a transitive 
> property. I think the fact that we are not seeing 
> (semanticRelation, TechnoPop) and (semanticRelation, 
> ElectroPop) might have to do with our FastClosure inference 
> program. I suspect that using the FullClosure program might 
> solve the latter problem. I would be surprised if it solved 
> the former too, but it is possible. Maybe give it a shot. 
> (See AbstractTripleStore.Options.CLOSURE_CLASS for details.)
> 
> Long story short, yes I think there is a bug. :)
> 
> I will log an issue for it in our bug tracker system.
> 
> If you are able to uncover the source of the problem please 
> let me know!
> 
> Thanks,
> Mike
> 
> 
> -----Original Message-----
> From: Jürgen Jakobitsch [mailto:jak...@pu...]
> Sent: Wednesday, April 28, 2010 1:52 AM
> To: big...@li...
> Subject: [Bigdata-developers] OWL Inference
> 
> hi, first of all congrats to this great project, i checked it 
> out once a couple of months ago and i can see that there some 
> significant improvements.
> 
> however might have found a bug in the inference engine, that 
> can be reproduced like so :
> 
> 1. instantiate a bigdata sail with fullfeatured properties 2. 
> add http://www.w3.org/2009/08/skos-reference/skos-owl1-dl.rdf 
> with baseURL http://www.w3.org/2004/02/skos/core
> 3. add http://turnguard.com/virtuoso/test10.rdf
> 
> now do the following sparql query
> 
> SELECT *
> WHERE {
> <http://www.turnguard.com/Music> ?p ?o
> }
> 
> the result from bigdata looks like this
> 
> http://www.w3.org/1999/02/22-rdf-syntax-ns#type
> http://www.w3.org/2004/02/skos/core#Concept
> http://www.w3.org/2000/01/rdf-schema#label "Music"@en 
> http://www.w3.org/2004/02/skos/core#narrower
> http://www.turnguard.com/ElectroPop
> http://www.w3.org/2004/02/skos/core#narrower
> http://www.turnguard.com/Pop
> http://www.w3.org/2004/02/skos/core#narrower
> http://www.turnguard.com/TechnoPop
> http://www.w3.org/2004/02/skos/core#narrowerTransitive
> http://www.turnguard.com/ElectroPop
> http://www.w3.org/2004/02/skos/core#narrowerTransitive
> http://www.turnguard.com/Pop
> http://www.w3.org/2004/02/skos/core#narrowerTransitive
> http://www.turnguard.com/TechnoPop
> http://www.w3.org/2004/02/skos/core#prefLabel "Music"@en 
> http://www.w3.org/2004/02/skos/core#semanticRelation
> http://www.turnguard.com/Pop
> http://www.w3.org/1999/02/22-rdf-syntax-ns#type
> http://www.w3.org/2000/01/rdf-schema#Resource
> 
> 
> the result from owlim (swiftowlim beta 12) like this
> 
> http://www.w3.org/2004/02/skos/core#narrowerTransitive
> http://www.turnguard.com/ElectroPop
> http://www.w3.org/2004/02/skos/core#narrowerTransitive
> http://www.turnguard.com/TechnoPop
> http://www.w3.org/2004/02/skos/core#narrowerTransitive
> http://www.turnguard.com/Pop
> http://www.w3.org/1999/02/22-rdf-syntax-ns#type
> http://www.w3.org/2004/02/skos/core#Concept
> http://www.w3.org/1999/02/22-rdf-syntax-ns#type _:node1 
> http://www.w3.org/2004/02/skos/core#semanticRelation
> http://www.turnguard.com/ElectroPop
> http://www.w3.org/2004/02/skos/core#semanticRelation
> http://www.turnguard.com/TechnoPop
> http://www.w3.org/2004/02/skos/core#semanticRelation
> http://www.turnguard.com/Pop
> http://www.w3.org/2004/02/skos/core#prefLabel "Music"@en 
> http://www.w3.org/2000/01/rdf-schema#label "Music"@en 
> http://www.w3.org/2004/02/skos/core#narrower
> http://www.turnguard.com/Pop
> 
> note that i'm positivly surprised by the inferred 
> rdf-schema#Resource with bigdata but there are significant 
> differences between the two inference engines, where owlim 
> seems to be right in most cases.
> 
> please note that :
> 
> in skos : narrower is a subProp of narrowerTransitive (same 
> for broader), both are inverse, the transitives are subProps 
> of semanticRelation, so the construct is a bit complex.
> 
> any suggestions if i'm doing some wrong or this is a bug 
> really welcome.
> 
> wkr www.turnguard.com/turnguard
> 
> -- punkt. netServices
> ______________________________ Jürgen Jakobitsch Codeography
> 
> Lerchenfelder Gürtel 43 Top 5/2
> A - 1160 Wien
> Tel.: 01 / 897 41 22 - 29
> Fax: 01 / 897 41 22 - 22
> 
> netServices http://www.punkt.at
> 
> 
> --------------------------------------------------------------
> ----------------
> _______________________________________________ 
> Bigdata-developers mailing list 
> Big...@li...
> https://lists.sourceforge.net/lists/listinfo/bigdata-developers
> 
> --
> punkt. netServices
> ______________________________
> Jürgen Jakobitsch
> Codeography
> 
> Lerchenfelder Gürtel 43 Top 5/2
> A - 1160 Wien
> Tel.: 01 / 897 41 22 - 29
> Fax: 01 / 897 41 22 - 22
> 
> netServices http://www.punkt.at
> 
> 
> --------------------------------------------------------------
> ----------------
> _______________________________________________
> Bigdata-developers mailing list
> Big...@li...
> https://lists.sourceforge.net/lists/listinfo/bigdata-developers
> 

Re: [Bigdata-developers] OWL Inference

Fast, scalable, robust graph database platform

Re: [Bigdata-developers] OWL Inference