一种基于改进的巴氏系数的协同过滤推荐算法

Collaborative filtering recommendation algorithm based on improved Bhattacharyya coefficient

传统基于邻居的协同过滤推荐方法必须完全依赖用户共同评分项,且存在极为稀疏的数据集中预测准确性不高的问题。巴氏系数协同过滤算法通过利用一对用户的所有评分项进行相似性度量,可以有效改善上述问题。但该种方法也存在两个很明显的缺陷,即未考虑两个用户评分项个数不同时的情况以及没有针对性地考虑用户偏好。在巴氏系数协同过滤算法的基础上进行了改进,既能充分利用用户的所有评分信息,又考虑到用户对项目的积极评分偏好。实验结果表明,改进的巴氏系数协同过滤算法在数据集上获得了更好的推荐结果,提高了推荐的准确度。

The traditional neighbor-based collaborative filtering recommendation method has to rely entirely on the common scoring items of users,and the accuracy of prediction in extremely sparse data sets is not high.Bhattacharyya coefficient collaborative filtering algorithm can effectively improve the above problems by using similarity measures for all the score items of a pair of users.But there are two obvious drawbacks to this approach,one is that it fails to consider the case that the number of scoring items of two users is not the same,the other is that there is no targeted consideration for user preferences.This paper improved the Bhattacharyya coefficient collaborative filtering algorithm,which could make full use of all the user’s rating information and consider the user’s positive rating preference for the project.Comparison of experimental results show that the improved Bhattacharyya coefficient collaborative filtering algorithm obtains better recommendation results on the dataset and improves the accuracy of the recommendation.