基于借阅时间评分的协同图书推荐模型与应用  

CF Recommending Model Based on Borrowing-time Scores and Its Application

作　　者：景民昌[1,2] 于迎辉[2] 

机构地区：[1]中国科学院研究生院,北京100080 [2]中国石油大学图书馆,北京102249

出　　处：《图书情报工作》2012年第3期117-120,共4页Library and Information Service

摘　　要：针对数字图书馆系统中的读者借阅历史数据,提出利用图书借阅时间计算读者兴趣度的评分数据模型。该模型把读者的借阅行为分为4种不同的类型:续借、超期借阅(长期超借与短期超借)、正常借阅、盲目借阅,设置阈值,分别计算其相对借阅时间,得出读者对图书的兴趣度。在该数据模型的基础上,实现一个基于协同过滤技术的图书推荐系统,验证模型的有效性。This paper proposes a new model for calculating interest level based on borrowing-time using readers＇ historical data in digital library system. In this model, the reader＇s borrowing book behaviors are divided into four types： renewing, extended borrowing, normal borrowing and blind borrowing, with different relative borrowing-time and thresholds to reflect their interest levels. At last, a book recommendation system is implemented based on collaborative filtering to verify the effectiveness of the model.

关 键 词：协同过滤 数据挖掘 图书推荐 用户兴趣度 

分 类 号：G250.7[文化科学—图书馆学]