Assessing the value of unrated items in collaborative filtering

In collaborative filtering systems, a common technique is default voting. Unknown ratings are filled with a default value to alleviate the sparsity of rating databases. We show that the choice of that default value represents an assump- tion about the underlying prediction algorithm and dataset. In this paper, we empirically analyze the effect of a vary- ing default value of unrated items on various memory-based collaborative rating prediction algorithms on different rat- ing corpora, in order to understand the assumptions these algorithms make about the rating database and to recom- mend default values for them.