摘要
为解决微博用户兴趣漂移问题,以人类记忆学中遗忘曲线为基础,提出一种微博用户兴趣模型,利用用户历史信息预测当前兴趣。在预测过程中,用户关注某信息的时间距离当前时间越远,该信息越容易被遗忘,其对用户当前兴趣的影响越小;用户关注某一领域的信息越多,印象越深刻,对该领域的兴趣度越高。这两点与人类对知识逐渐遗忘和重复学习的过程具有高度相似性,因此该模型预测准确性更高。实验结果表明,该模型能较好地预测微博用户兴趣,召回率可达85.3%,实用性较强。
To solve the problem of the micro-blog user interest drift ,a micro-blog user interest model based on the forgetting curve was presented .The current interest was predicted by the history information of users .In the predicting process ,the longer the time from the users’ attention for a message to the current ,the weaker the influence of the message ;and the higher interest degree to the field ,the more attention users paid to a concerned field .These two points are regarded as the process of human gradually forgetting and repeatedly learning knowledge .Therefore ,the model possesses a higher accuracy .Experimental results show that the model can predict the micro-blog user interest better with the recall rate of 85.3% and good practicality .
出处
《计算机工程与设计》
CSCD
北大核心
2014年第10期3367-3372,3379,共7页
Computer Engineering and Design
基金
国家863高技术研究发展计划基金项目(2011AA010603
2011AA010605)
关键词
微博
预测
用户兴趣
重复学习
遗忘曲线
micro-blog
predict
user interest
repeated learning
forgetting curve