目前电子商务推荐系统中存在客户评分稀疏性问题,隐性评分是解决该问题的有效方法,但现有方法只考虑客户对单个网页的兴趣度.客户浏览行为和网页之间语义相似度可以综合客观反映客户对商品隐性的评分.建立网页商品概念-属性矩阵CA(Goods Concept-Attribute Matrix),综合考虑商品属性、相关商品及其在网页中的分布等因素,基于客户浏览路径和时间的统计分析,通过语义词林WordNet计算网页语义相似度,综合计算商品语义隐性评分.通过算例和实证研究说明该算法的有效性. At present, the e-commerce recommendation system has the problem of customer scoring sparseness, implicit scoring is an effective way to solve the problem, but the existing method only considers the customer’s interest in a single web page. The semantic similarity between customer browsing behavior and web page can be integrated Objectively reflect the customer’s implicit score on the product.Create the concept of goods matrix-attribute matrix (CA), considering the attributes of goods, the related products and their distribution in the webpage, based on the customer’s browsing path and time Through the semantic word forest WordNet, the semantic similarity of the web pages is computed, and the semantic recessive score of the commodity is calculated synthetically.The effectiveness of this algorithm is illustrated by examples and empirical studies.