Raw

Now this is more like it: Jon Udell: Collaborative filtering with del.icio.us

The prototype is a two-stroke engine. First, it captures the set of del.icio.us users on the steep part of the curve - the ones with whom I have the most link affinity. Then it reads all their RSS feeds, coalesces the links, and applies another filter to select just the links above a threshold of commonality.

He also mentions combining different data handling methods (recommending book " Data Crunching"). Yep. In particular for something like this, rather than having a load of different standalone apps apps, it would make sense to combine the affinity metrics. Because it's not clear what systems you might want to combine the data with in advance (movie recommenders? social networking? spam filters?) , it would make sense to express the data declaratively, systematically. I reckon that has to mean RDF.