A Technique of Recursive Reliability-Based Missing Data Imputation for Collaborative Filtering

Collaborative filtering (CF) is a recommendation technique that analyzes the behavior of various users and recommends items preferred by with similar preferences. However, CF methods suffer from poor accuracy when user preference data used in process sparse. Data imputation can alleviate sparsity problem substituting virtual part missing In this paper, we propose k-recursive reliability-based (k-RRI) first selects high reliability then recursively imputes additional selection while gradually lowering criterion. We also new similarity measure weights common interests indifferences between items. The proposed method overcome disregarding importance resolve existing methods. experimental results demonstrate approach significantly improves compared to those resulting state-of-the-art demanding less computational complexity.