An Improved Collaborative Method for Recommendation and Rating Prediction

User-Item matrix (UI matrix) has been widely used in recommendation systems for data representation. However, as the amount of users and items increases, UI matrix becomes very sparse, which leads to unsatisfactory performance in traditional recommendation algorithms. To address this problem, in this paper, a rating prediction method with low sensitivity to sparse datasets is proposed. This method incorporates tag information and factor analysis approach that has been successfully applied in various areas, to discover the most similar top-N users based on the similarity of users' inner idiosyncrasies. Based on the most similar top-N users discovered, an improved collaborative filtering method is designed for rating prediction and recommendation. Extensive experiments have been done for comparing the proposed method with traditional collaborative filtering and the matrix factorization methods. The results demonstrate that our proposed method can achieve better accuracy, and it is less sensitive to sparseness of datasets.