I thought I would write a quick post to update people on recent work I have been doing in improving dotNetRDF for the next release. One of the major changes in the new version is a new internal storage mechanism for Triples which now uses a full blown Hash Table which makes storing colliding Triples much easier. Hash Tables are also used for creating a number of indexes to significantly improve the speed of some types of lookups. This change comes with a minor performance impact on loading times and increased storage requirements but the benefits are immediately clear in our benchmarking, in particular this change gives much better SPARQL performance and a more graceful scaling of performance with number of Triples (appears linear rather than exponential).
One of the main things I've been working on is implementing various aspects of the SPARQL 1.1 working draft which seem most stable into the SPARQL Engine. So far I have implemented Project Expressions, Aggregates along with simple GROUP BY and HAVING clauses.
In addition to this I have now added support for 4store into the library which makes 3 different native Stores that can be used out of the box - Talis Platform, Virtuoso Universal Server and 4store.
At the moment it looks like I will probably do the 0.1.2 Alpha Release at the end of November depending on other time constraints. 11/11/2009 12:03:17 by Rob Vesse in English
There are currently no Tags for this Content!