摘要
针对没有考虑用户和项目的相似性,推荐精度较低的问题,提出评分统计预测模型,定义用户和项目统计信息分别表示用户个人喜好和项目品质;根据统计信息,产生符合正态分布随机数来表示项目品质评分和个人喜好的评分;最后结合两个评分建立线性回归预测模型进行评分预测,并据此设计了相关算法。实验结果表明,文章提出的算法推荐精度高于传统协同过滤算法。
In this paper, a statistical prediction of ratings model is proposed, in order to solve the problem that the traditional collaborative filtering algorithm based on prediction of ratings is not accurate enough in prediction since it fails to consider the similarity of users or items. User statistical information and item statistical information are defined to denote the user preferences and the quality of item respectively. Statistical information is used to generate random data in accordance with Gauss distribution, which can represent the prediction ratings of user preferences and item quality. Final- ly, a linear regression prediction model combining the two prediction ratings is put forward to make predictions. Based on this model, the relative algorithm is designed. The experimental results show that the proposed algorithm performs better than traditional algorithms.
出处
《信息工程大学学报》
2016年第6期646-650,693,共6页
Journal of Information Engineering University
基金
国家自然科学基金资助项目(61572520)
关键词
协同过滤
评分统计预测
用户统计信息
项目统计信息
collaborative filtering
statistical prediction of ratings
user statistical information
item statistical information
