B-Rank: A top N Recommendation Algorithm

In this paper I propose B-Rank, an efficient ranking algorithm for recommender systems. B-Rank is based on a random walk model on hypergraphs. Depending on the setup, B-Rank outperforms other state of the art algorithms in terms of precision, recall ∼ (19%− 50%) and inter list diversity ∼ (20%− 60%). B-Rank captures well the difference between popular and niche objects. The proposed algorithm produces very promising results for sparse and dense voting matrices. Furthermore, I introduce a recommendation list update algorithm to cope with new votes. This technique significantly reduces computational complexity. The algorithm implementation is simple, since B-Rank needs no parameter tuning.