提出一种新的基于双因素方差分析的推荐算法DFAR。该方法基于成熟的统计学模型,简单易理解,具有很好的鲁棒性。实验结果证明,该算法相比传统的项目协作过滤算法取得了更好的推荐效果,并大大节省了算法所需要的空间。
This paper presented a recommendation algorithm, named DFAR, which was based on double-factorial analysis of variance. DFAR had well studied statistical properties, easy-intelligibility and good robusticity. Experiments show that this algorithm can achieve better prediction accuracy than traditional item-based CF algorithm. Furthermore,it can dramatically save the storage space.