LDOS CAMRA 2011 submission

<p>This paper is a report on the work done by the LDOS<br>team (from the University of Ljubljana Faculty of electri-<br>cal engineering) on the 2011 RecSys CAMRA (Challenge on<br>Context-aware Movie Recommendation). We present three<br>approaches in the track 1 competition which requires to se-<br>lect the top N recommended items for each household. Our<br>general approach uses a matrix factorization algorithm to<br>compute per-user rating predictions. The context is taken<br>into account using an averaged and weighted sum to cal-<br>culate the household-based rating predictions. We also ex-<br>plored how the limiting of true positive candidates with ad-<br>ditional knowledge from the dataset influences the perfor-<br>mance of our models.</p>