摘要
在信息爆炸的时代,推荐系统作为重要的工具可以帮助人们从众多的数据中迅速筛选出想要的信息。由于用于预测分数的Resnick公式只关注了用户的评分记录而没有关注用户评分之间的影响,因此提出使用BP神经网络代替Resnick公式进行评分预测,同时针对推荐系统中存在的数据稀疏问题,提出采用SVD矩阵分解来填充用户项目矩阵,最终得到改进的混合推荐算法。通过用标准Movie Lens数据集对改进的混合推荐算法与传统的协同过滤进行对比,最终数据显示改进的混合推荐算法更加准确。
In the era of information explosion,recommendation system as an important tool can help people quickly screen out the desired information from numerous data.Because the Resnick formula used to predict scores only focuses on the user's rating records,but does not pay attention to the influence between us er's scores.Therefore,BP neural network is proposed to replace Resnick formula for score prediction.At the same time,aiming at the problem of data sparsity in the recommendation system,SVD matrix decomposition is used to fill the user item matrix,and an improved hybrid recommendation algorithm is finally obtained.By using standard Movie Lens dataset to compare the improved hybrid recommendation algorithm with the traditional collaborative filtering algorithm,the final data show that the improved hybrid recommendation algorithm is more accurate.
作者
孙乐
Sun Le(College of Computer Science and Engineering,Anhui University of Science&Technology,Huainan Anhui 232001,China)
出处
《信息与电脑》
2020年第21期41-43,共3页
Information & Computer
关键词
推荐算法
BP神经网络
SVD矩阵分解
算法改进
recommendation algorithm
BP neural network
SVD matrix factorization
algorithm improvement
