面向用户偏好的属性值评分分布协同过滤算法

Collaborative filtering algorithm based on rating distribution of attributes faced user preference

针对传统协同过滤算法存在的不足,本文充分考虑用户对项目相关属性特征的偏好,将用户对项目的评价转化为用户对项目属性偏好的评分分布;在此基础上,对传统的协同过滤算法的相似性度量方法进行改进,并采用修正的用户偏好数学期望预测模型,提出一种面向用户偏好的属性值评分分布协同过滤推荐算法.实验结果表明,该算法可有效解决传统过滤算法存在的问题,推荐精度显著提高,使推荐服务更好地满足用户的偏好需求.

Considering the related attribute features of user s preference,a novel collaborative filtering algorithms is proposed,which is based on rating distribution of attributes.In new algorithm,the user s rating to an item is converted to user s rating distribution of attributes to the item.The traditional similarity measure method is developed,and the amendatory prediction model for mathematical expectation is introduced.Experiment on typical data set show that the proposed algorithm can provide excellent recomm...