A dynamic collaboration filtering recommendation approach based on Hidden Markov Model (HMM)and DBN is proposed. The approach to collaboration filtering recommendation based on Hidden Markov Model simulates user' s behaviors while a user is browsing Web pages, and sets up the nearest-neighbor set on his behaviors.Then the DBN recommendation model is constructed on this method. The model to update recommendation model is used when the new data are added. An experiment shows the excellent performance of his approach.%提出了一种基于动态贝叶斯网络的隐马尔可夫协同过滤推荐的新方法.基于隐马尔可夫模型的协同过滤方法模拟用户在浏览网页时的行为,根据用户浏览网页时的行为建立最近邻集合.在基于隐马尔可夫协同过滤推荐技术的基础上,构造基于DBN的推荐模型.当有新类型的数据加入时,用此模型来更新推荐模型.实验表明,此方法具有较高的推荐质量.
展开▼
