Enhancement of collaborative filtering performance under data scarcity environment

This study devotes to improve the prediction accuracy of prediction algorithms in recommender systems which one is collaborative filtering algorithm to estimate user's preference to items transacted on the web. From the our experiment, data scarcity problem is critical factor for decreasing prediction accuracy so the method for reducing data scarcity is meaningful way to increase prediction accuracy and also techniques for improving prediction accuracy like as the significant weight must be applied to the prediction process. This study proposes substitution methods like as the means of modes of users and items are effective and economical ways to reduce data scarcity better than other complicate substitution methods and can get even more accurate result than original result.