Katie and I were at the Jena User Conference in Bristol on the 10th and 11th of May. The conference had a good mix of papers and demos that covered a variety of subjects. We met a number of people and got useful pointers and ideas from them. There were some papers that we found particularly useful and are keen to try out ourselves in the Metastore project.

One of them was Chris Dollin's Eyeball tool for validation of the data in the triplestore. This would be great for sanity-checking the metastore data given the extreme flexibility of the store. The main question would be whether we can run it for the huge amount of data in the store and if yes, then the amount of time it would take to run it. I think we will try this on a single-publisher store initially and then on increasing sizes of the store...certainly something to look into further.

Another was Max Völkel's RDFReactor - it transforms an ontology in RDF Schema into a Java object model. In the Metastore project, we had to go through a number of iterations and a lot of head-scratching while translating our schemas to a Java object model that would most closely follow the schemas. So it would be interesting to have a look at the model that RDFReactor generates.

An interesting development was the Jena property tables by Kevin Wilkinson to store patterns of RDF statements.

Another paper that I liked was Kate Byrne's Tethering Cultural Data with RDF, particularly the natural language processing part- something I would like to explore in my spare time (what there is of it!)

We presented our paper "Scaling Jena in a commercial environment: the Ingenta Metastore Project" and got positive feedback and suggestions. A few people were interested in the customised schemas that we have developed for representing some parts of our dataset.

Max Völkel was interested in using our schemas to compare the object models generated automatically by RDFReactor with the ones that we have developed ourselves.

In the paper, we had mentioned the out-of-memory problems we had with using OWL for inferring relations, due to the sheer size of our database. The Jena developers gave us some useful suggestions on things that we could try to make this work and also asked us for examples of relations that we would want to infer.

The conference was a very enjoyable learning experience for us - we even managed to present our paper without too much panic or too many mishaps :-)

Thanks to the Jena team for organising such a great event...looking forward to the 2007 conference!