基于读者借阅二分网络的图书可推荐质量测度方法及个性化图书推荐服务

The Measures of Books' Recommending Quality and Personalized Book Recommendation Service Based on Bipartite Network of Readers and Books' Lending Relationship

本文首先提出一种利用读者借阅行为特征来判断图书可推荐质量的思路,并结合读者图书借阅关系所形成的二分网络结构,设计了一种测度图书可推荐质量的迭代算法,从而为个性化图书推荐服务提供了良好的推荐客体。在上述研究的基础上,结合图书类别目录层次、标题语义信息的提取处理方法、基于加权XML模型的用户个性化模式表达方法及其权值扩散策略,提出了三种图书馆个性化图书推荐服务的形式,分别是特定主题的图书推荐服务、现有所借图书的修正型推荐服务和新书推荐服务。最后,文章对相关测试实验及其效果做了必要的说明。图9。表10。参考文献13。

This paper begins with the introduction of an idea about judging books＇ recommending quality through the charac- teristic analysis of the readers＇ lending behavior. Based on the bipartite network structure of reader and books＇ lending relation- ship, an iterative algorithm which can recognize the book＇s recommending quality is proposed. This algorithm can provide better recommendation objects for personalized book recommendation service. On the basis of the research mentioned-above, combined with the hierarchy of book catalog, the extract of semantic information of titles, the expression of personalized user profile based on weighted XML model and weight＇s spreading strategy, th/s paper also discusses three diffe^nt methods of personalized book recommendation service in library, which include recommendation service of specific topic books, revised recommendation serv- ice of harrowed books, new books recommendation service. Finally, this paper reports some related experiments which show the improvement of effect. 9 figs. 10 tabs. 13 refs.