用户兴趣模型的更新与遗忘机制研究

Research On the Update and Forgetting Mechanism of User's Interesting Model

根据HTML文档的特点,赋予不同标签限定的特征词相应的权重系数,从而计算特征词的权值,并根据用户的浏览速度更新兴趣模型,其次,采用在在用户模型的特征向量中引入该特征词最后一次的更新时间,并结合遗忘因子修正特征词的权重的方法,实现了模型的遗忘。实验证明,引入遗忘机制的个性化推荐系统获得了较高的推荐效率。

For semi-structured HTML document features, given different feature words corresponding weight coefficient which be limited with labels, inorder to calculate the weight of feature word, and update the user model according to the users' browsing speed, secondly, introduce the last updated timeof the feature word in the user model vectors, combining the forgetting factor to revise the weight of feature word, so achieving the model forgotten, andrealize the higher recommendation efficiency of the system.