| Abstract |
In the half hour, I would like to discuss two ideas that are relevant for those who work
with RDF. First, an idea about identifying video content which is inspired by the FRBR model used in the library domain. This model takes a layered approach, where an original work "Bezorgde ouders" (from Gerard Reve) can have several expressions, like an english translation or a censored version. It can have different manifestations, e.g. a paperback version, 2nd press. And at last several items, which is the book in my bookshelf, or the PDF downloaded via a torrent. These distinctions need different identifiers, however are related. This is relevant for recommender algorithms and 'more info' functionality. Same for vidieo content.
The second topic, is a proposal for simple access policies for RDF databases. The idea is closely related to the work on warrants from Jeremy Carroll and Christian Bizer, but much simpler and directly compatible with any RDF store that understands SPARQL. Simply put, a contract is a simple sparql construct query in combination with the allowed activity (ADD, REMOVE, QUERY). The contract is processed via cronjobs and the results are placed in a Bloomfilter. When at query time the user queries, the result set is only returned (or deleted) when the result triples are in the bloomfilter (otherwise the user violated the contract). "Addition", is a bit more complicated.
|