Hello Tobias,

On Wed, 2012-05-30 at 12:43 +0200, Tobias Neef wrote:
> - how is the delta calculated?
> - Is this feasible for a large RDF document/database?


Virtuoso's delta calculator works with different speed depending on the
structure of data in graphs and the known semantics. With only IRIs in
data and no blank nodes, it's quite fast. Blank nodes make matching two
graphs more complicated and the quality of the diff vary depending on
accuracy of matching individual bnodes by their functional and inverse
functional properties (FP and IFP). Without sufficient amount of these
properties, the diff may grow beyond practical limits. The speed of diff
with bnodes can be tweaked, to get big diff faster or to get small diff
slower.

> - Is there only a way to get notifications for a complete document or
> also for subgraphs?

I don't know it personally, sorry, someone else will clarify.

Best Regards,

Ivan Mikhailov
OpenLink Software
http://virtuoso.openlinksw.com

P.S., some Virtuoso-based PubSubHub servers may provide a web page
named /GraphCmp/, to make experiments with diff logic used at that
servers.



Reply via email to