摘要
在读者向图书馆借阅图书或从书店购买图书的过程中,名人推荐及畅销书榜单对读者的选择具有很大的影响。针对这种情况,结合影响力分析和主题模型提出新的图书协同过滤推荐系统。算法结合最大熵和最大方差来选择评分矩阵的影响力用户和影响力项目,基于建立的密集矩阵预测未知评分。运用改进的聚类算法对词向量进行聚类处理,建立主题。在公开的数据集上完成验证实验,结果表明该算法提高了图书推荐系统的性能。
When readers borrow books from libraries or purchase books from shops,both celebrity recommended books and best seller list have a big influence to the selection of readers.In view of this,we propose a new book collaborative filtering recommendation system,which combines influence analysis and topic model.This algorithm combined maximum entropy and maximum variance to select the influential users and influential items in the rating matrix,and it predicted the unknown ratings based on the dense matrix.The algorithm applied enhanced clustering algorithm to cluster the word vectors,as a result,the topics of texts were constructed.Validation experiments were carried on public datasets.The results show that the proposed algorithm improves the performance of book recommendation system.
作者
成胤钟
Cheng Yinzhong(Chongqing Vocational College of Culture and Arts,Chongqing 400067,China)
出处
《计算机应用与软件》
北大核心
2024年第1期64-70,104,共8页
Computer Applications and Software
关键词
图书推荐系统
主题模型
球面k均值聚类
最大熵
协同过滤
意见领袖
Book recommendation system
Topic model
Spherical k-means clustering
Maximum entropy
Collaborative filtering
Opinion leader
