摘要
构建图书推荐系统,不仅可以让用户快速有效地获取所需图书信息,减少信息过载,同时也可以较好地发挥图书馆馆藏图书资源的潜在价值。在综述了几种常用推荐算法的基础上,给出了基于协同过滤的推荐算法的实现过程,并针对其冷启动和数据稀疏性问题给出了优化方案及优化后的算法实现流程。结果表明:在算法中引入用户特征属性与用户聚类方法,有效降低了数据稀疏性问题,提升了算法的推荐效率,一定程度上解决了图书推荐系统中的推荐算法设计。可以将该优化后的算法运用于图书馆的图书推荐系统设计中。
Constructing a book recommendation system can not only allow users to quickly and effectively obtain the required book information and reduce information overload,but also can make better use of the potential value of the library’s collection of book resources.On the basis of summarizing several common recommendation algorithms,the implementation process of recommendation algorithm based on collaborative filtering is given,and the optimization scheme and the optimized algorithm implementation process are given for its cold start and data sparsity.The results show that the user characteristic attribute and user clustering method were introduced into the algorithm,which effectively reduced the data sparsity problem,improved the recommendation efficiency of the algorithm,and solved the recommendation algorithm design in the book recommendation system to a certain extent.The optimized algorithm can be used in the design of library book recommendation system.
作者
张德青
程锦
ZHANG Deqing;CHENG Jin(School of Computer Engineering Anhui Sanlian University,Hefei Anhui 230601,China)
出处
《西昌学院学报（自然科学版）》
2021年第2期78-81,共4页
Journal of Xichang University(Natural Science Edition)
基金
安徽三联学院校级自然科学基金项目(KJYB2019008)
安徽省教育厅自然科学基金重点项目(KJ2019A0891)。
关键词
推荐算法
相似度
数据稀疏
recommendation algorithm
similarity
data sparsity
