Item-Based Collaborative Filtering Recommendation Algorithm Combining Item Category with Interestingness Measure

Abstract

In order to overcome the limitations of data sparsity and inaccurate similarity in personalized recommendation systems, a new collaborative filtering recommendation algorithm by using items categories similarity and interestingness measure is proposed. In this algorithm, first the items categories similarity matrix is constructed by calculating the item-item category distance, and then analyzes the correlation degree of different items by using interestingness measure, last an improved collaborative filtering algorithm is proposed by combining the information of items categories with item-item interestingness and utilizing improved conditional probability method as the standard item-item similarity measure. Experimental results show this algorithm can effectively alleviate the dataset sparsity problem and achieve better prediction accuracy compared to other well-performing collaborative filtering algorithms.