中文摘要：

个性化推荐技术能够帮助用户更方便地从大量的文本数据中得到感兴趣的文本.数字图书馆中现有的个性化推荐技术都是根据文本相似性为用户推荐感兴趣的文本.该文提出用户对文本的兴趣度的概念,综合考虑了文本之间的相似性、文本的信息量和新颖性3个因素,比相似性能更好地反映用户的兴趣.同时提出基于兴趣度的个性化推荐算法.理论分析和实验结果均表明,基于兴趣度的推荐算法的推荐完全性和准确性比相似性推荐算法和基于图的混合推荐算法均有显著提高.

英文摘要：

Personalized recommendation techniques can actively provide users with useful information gleaned from massive and otherwise unmanageable resources. Traditional recommendation algorithms in digital libraries are based on textual similarities between documents. This paper gives a definition of degree of interest. It considers three factors： textual similarities between documents, informational volume, and informational novelty. This reflects user＇s interest better than textual similarity alone. An effective interest-based recommendation algorithm was then proposed. Theoretical analysis and experimental evaluations demonstrated that the recommendation algorithm based on the degree of interest can improve both the completeness and accuracy of recommendations when compared to recommendation algorithms based on similarity and mixed recommendation algorithms based on graph.