Sparse Online Learning for Collaborative Filtering

With the rapid growth of Internet information, our individual processing capacity has become over-whelming. Thus, we really need recommender systems to provide us with items online in real time. In reality, a user's interest and an item's popularity are always changing over time. Therefore, recommendation approaches should take such changes into consideration. In this paper, we propose two approaches, i.e., First Order Sparse Collaborative Filtering (SOCFI) and Second Order Sparse Online Collaborative Filtering (SOCFII), to deal with the user-item ratings for online collaborative filtering. We conduct some experiments on such real data sets as MovieLens100K and MovieLens1M, to evaluate our proposed methods. The results show that, our proposed approach is able to effectively online update the recommendation model from a sequence of rating observation. And in terms of RMSE, our proposed approach outperforms other baseline methods.