If generating RDF, what is the best order for loader performance?

Virtuoso Universal Server Tips, Tricks, and HowTos
1

For index, and compression efficiency it can be interesting to presort data before loading. Is true for virtuoso? If so for the RDF data what is normally the most effective sorted order?

PSOG was my guess if there are limited predicates in use (order’s of magnitude less distinct predicates than subjects and objects) and only a handful of graphs. Or would GPSO be better in such a case?

Or is it a case of test it and see :wink:

2

The pre-sorting of a dataset on P (predicate) attribute may have better loading speed as the 2 full RDF indexes are on a leading P. We do not expect a big difference however as the RDF storage is a complex structure where we have RDF objects, IRIs, datatypes, languages etc, like for example : <#s> <#name> “cat”@en, “chat”@fr .