[redland-dev] raptor turtle serializer scalability patch

John Fieber jfieber at adobe.com
Mon Nov 16 19:18:10 CET 2009


On Nov 16, 2009, at 9:20 AM, Dave Beckett wrote:

> I've had a bit of a look at this and it seems the rdfdiff 
> rdf-graph-compare algorithm is naive/broken and relies on the semantics of 
> the current raptor_abbrev (rdfxml-abbrev and turtle serializing) module 
> too much.  I'm not sure what I'll do about this but I'll carry on looking.
> 
> At this stage I'd like to commit the patch (with some minor code style 
> fixes) but I would really like to get rdfdiff working first.  Hmm.

I have a collection of tests marked "pending RDF graph isomorphism check", waiting for some time to implement.  Some references...

http://www.hpl.hp.com/techreports/2003/HPL-2003-142.html
http://www.hpl.hp.com/techreports/2001/HPL-2001-293.pdf

And one other that is evading me that clearly outlines the isomorphism algorithm shortcuts you can make based on the specifics of RDF graphs.  Jena has an implementation.   This may have some useful information, but I haven't read it:

https://www.ifi.uzh.ch/fileadmin/site/teaching/Diplomarbeiten/Abgeschlossene_Diplomarbeiten/Jahrgang_2006/Baggenstos_Daniel.pdf

Of course, if it were not for blank nodes, this would be trivial.

-john



More information about the redland-dev mailing list