Abstract:

In this paper, a novel idea is proposed to observe the shifts of college readers’ reading preferences over time based on the records of readers’ borrowing behavior and the information of corresponding books. With the help of the characteristics of the bipartite network, a calculation method is proposed for measuring the characteristics of preference shift and combining it with traditional book recommendation as an object algorithm hybrid recommendation. At the same time, using the historical data of book borrowing from the library in Shanghai Jiao Tong University, the superiority of the calculation results of preference changes and the hybrid recommendation method combining preference changes are demonstrated, compared with the traditional recommendation method.

Key words: preference shift, book recommendation, personalized recommendation, academic library

摘要：

提出了一种根据读者借阅行为记录和图书信息来观察高校读者阅读偏好随着时间变化的思路，并借助二分网络结构特性，提出了测量偏好变化特征的计算方式及其作为推荐客体结合传统图书推荐算法的混合推荐方式。同时，使用上海交通大学图书馆图书借阅历史数据展示了偏好变化计算结果和结合偏好变化的混合推荐方法相比传统推荐方法的优越性。

关键词: 偏 好变化, 图书推荐, 个性化推荐, 高校图书馆