提出了一种Web用户兴趣变化识别的算法,面向层次结构的信息类网站,可以达到客户自适应的目的.该算法由长、短期两个兴趣模型构成,分别基于指数衰减理论和贝叶斯后验概率理论.基于一个Internet上真实网站的实验结果表明,本文提出的算法可以迅速识别出用户的兴趣变化,并且广泛适用于新闻、虚拟社区等层次结构类网站的网页内容推荐,满足用户在信息浏览时的个性化需求. This paper proposes an algorithm for changing the interest of Web users, which is oriented to the information-type website with a hierarchical structure and can achieve the purpose of customer self-adaptation.The algorithm consists of long and short interest models based on exponential decay theory and Bayesian Experimental Probability Theory Based on the experimental results of a real website on the Internet, the proposed algorithm can quickly identify changes in user interest and is widely applicable to the web content recommendation of hierarchical websites such as news and virtual communities to meet the needs of users in information Personalized needs when browsing.