摘要
个性化信息检索系统的实时性关键在于如何动态更新用户兴趣模型。针对原有方法的不足,改进用户兴趣模型的描述与更新方式。首先根据网页文档的特征改进TF-IDF(Term Frequency-Inverse Document Frequency)算法,以此作为用户兴趣特征词的权重,同时通过引入领域本体,将用户兴趣特征项进行语义扩展,并根据用户浏览行为,改进其用户兴趣主题计算方式,并在此基础上提出用户兴趣模型的更新与遗忘机制。实验对比结果表明,该方法能够捕捉用户兴趣的变化,进一步提高个性化信息检索的准确度与用户满意度。
It is essential for the real-timeliness of personalized information retrieval system how to dynamically update the user interest model. Aiming at the deficiency of the existing method, the paper improves the user interest model description and updating mode. Firstly, ac- cording to the characteristics of the web document, TF-IDF algorithm is improved, whose results are taken as user interest feature words' weight;meanwhile, by introducing domain ontology, the user interest feature items are semantically extended;then, according to user browsing behavior, the user interest theme calculation method is improved, based on which the updating and forgetting mechanism of user interest mod- el is proposed. Results from comparative experiments indicate that the method can capture changes of the user interest and further improve the precision of personalized information retrieval and user satisfaction.
出处
《计算机应用与软件》
CSCD
北大核心
2014年第3期7-10,共4页
Computer Applications and Software
基金
兰州文理学院科研能力提升计划项目(2013YBTS03)