Improvement of Personalized Recommendation Algorithm Based Content-boosted Collaborative Filtering Algorithm

Collaborative filtering (CF) systems have been proven to be very effective for personalized and accurate recommendations. These systems are based on the Recommendations of previous ratings by various users and products. Since the present database is very sparse, the missing values are considered first and based on that, a complete prediction dataset are made. In this paper, some standard computational techniques are applied within the framework of Content-boosted collaborative filtering with imputational rating data to evaluate and produce CF predictions. The Content-boosted collaborative filtering algorithm uses either naive Bayes or means imputation, depending on the sparsity of the original CF rating dataset. Results are presented and shown that this approach performs better than a traditional content-based predictor and collaborative filters.