Ideas for 2007 — tentative

We should do anything that got cancelled in 2006. Ricardo expressed interest in revisiting relational learning. There seemed to be a fair amount of interest in time series modelling, possibly interaction with CS on this?