SemanticMetadata.net

Report from the Semantic Web Atelier

October 19, 2006 on 1:27 pm | | In Dev, General, SemanticWeb, Software | Comments Off

Yesterday the 9th SWAt (Semantic Web Atelier) in Graz took place. As I would write a summary anyway I’m blogging it here and now:We two talks, the first one was from Peter Scheir about the performance of RDF databases and the second one was from Gisela Dösinger about the findings of a survey.

The RDF databases part was in general a lively discussion as there are not many good scientific evaluation results available on the internet. It is true for all databases that were tested (Sesame, Jena, Kowari, …) that inference and reasoning kills performance. Furthermore if the API instead of query languages for operations on the data is used the performance is much better. In general Sesame is faster than Jena, only performance of delete operation is dreadful within Sesame. The Sesame API and the processes inside suffer from missing documentation. Jena on the other side has enough docs and is therefore easier to use. SPARQL support is not very good in both cases. One essence of the discussion was that if it is possible (based on the use case) one should use a database, like for instance MySQL, for storing and retrieving the triples.

Gisela Dösinger presented the results of an interview and online survey done within the Sembase project. Many people from Austria, who are somewhat involved with IT and / or Semantic Web were questioned. One major finding can be summarized as follows: The interview results show for example that the Semantic Web lacks clear definitions of problems, results and benefits as well as best practice examples. More results and corresponding figures are available the presentation slides.

The slides of both presentations will be available here.