基于隐式反馈的用户兴趣漂移方法  

METHOD OF DRIFTING USER’S INTERESTS BASED ON IMPLICIT FEEDBACK

作　　者：伍大清[1] 阳小华[1] 马家宇[1] 胡东[2] 吴取劲[1] 

机构地区：[1]南华大学计算机科学与技术学院,湖南衡阳421001 [2]北海舰队驻南华大学选培办,湖南衡阳421001

出　　处：《计算机应用与软件》2010年第9期88-90,共3页Computer Applications and Software

基　　金：湖南省科技厅2006年计划项目(2006GK3086)

摘　　要：针对现有用户兴趣模型在模型建立以及更新阶段漂移策略的缺陷,设计了一种改进的基于隐式反馈的用户兴趣漂移方法。将用户兴趣分为长期兴趣和短期兴趣,对短期兴趣采用滑动窗口处理更新,对长期兴趣采用最近最少使用淘汰算法,并将该模型与遗忘策略模型、滑动时间窗口模型做了精确率的对比实验,实验结果显示该模型的性能优于其它三个模型。In the paper we designed an improved user＇ s interests drifting approach based on implicit feedback for dealing with the＂ limita- tion of excursion policy in building and updating step of user＇ s interest model. In this model, user＇ s interests consist of long-term and short- term. For short-term interests we use sliding window algorithm to deal with its updating. For long-term interests the gradual forgetting algorithm based on least recently used algorithm is used. Contrast experiments of accuracy were made on the model and the gradual forgetting policy model as well as the sliding time window model, the results showed that the performance of this model is better than other three models.

关 键 词：用户兴趣模型 VSM 遗忘策略 滑动时间窗口 

分 类 号：TP393.4[自动化与计算机技术—计算机应用技术]