针对如何在邮件系统中进行用户兴趣模型更新的问题,提出了一种兴趣漂移方法。根据用户长期兴趣和短期兴趣的不同特点,对短期兴趣模型采用时间窗口方法,对长期兴趣模型采用遗忘函数方法进行模型更新。实验结果表明,短期兴趣模型能很好地预测用户近期的兴趣,长期兴趣模型对较长时间段内用户兴趣的预测更稳定。采用混合模型能更好地描述用户兴趣。
Aiming at the problem of how to update users' interest models in the E-mail system,this paper provides a method of interest drifting.According to the difference between the long-term interest and the short-term interest,the time window algorithm is used in the short-term interest model,and the forgetting function algorithm is used in the long-term interest model.The experiment results demon-strate that the short-term interest model can well predict the user's recent interests.And the long-term interest model is more stable for a long time.The hybrid model can better describe user's interests.