A Hybrid Recommendation Method with Reduced Data for Large-Scale Application

Most recommendation algorithms attempt to alleviate information overload by identifying which items a user will find worthwhile. Content-based (CB) filtering uses the features of items, whereas collaborative filtering (CF) relies on the opinions of similar customers to recommend items. In addition to these techniques, hybrid methods have also been suggested to improve the performance of recommendation algorithms. However, even though recent hybrid methods have helped to avoid certain limitations of CB and CF, scalability and sparsity are still major problems in large-scale recommendation systems. In order to overcome these problems, this paper proposes a novel hybrid recommendation algorithm HYRED, which combines CF using the modified Pearson's binary correlation coefficients with CB filtering using the generalized distance-to-boundary-based rating. In the proposed recommendation system, the nearest and farthest neighbors of a target customer are utilized to yield a reduced dataset of useful information by avoiding scalability and sparsity problem when confronted by tremendous volumes of data. The use of reduced datasets enables us not only to lessen the computing effort, but also to improve the performance of recommendations. In addition, a generalized method to combine CF and CB system into a hybrid recommendation system is proposed by developing on the normalization metric. We have used this HYRED algorithm to experiment with all possible combination of CF and statistical-learning-based CB filtering. These experiments have shown that the use of reduced datasets saves computational time, and neighbor information improves performance.