中文摘要：

商品属性和用户浏览商品的记录是进行商品推送的关键。对商品属性进行了定义并给出了针对商品互动（行为）信息的相似度计算公式,进而提出了一种商品推荐的改进slope one协同过滤算法。在该算法中提出基于商品名称的二进制串进行商品间名称型属性海明距离的度量依据,并基于某电子商务平台站点提供的商品信息、客户信息、客户对商品的浏览记录以及商品的稀疏度等信息,验证了所提算法的有效性。通过与基本slope one算法和加权slope one算法的比较,证明了算法的合理性。

英文摘要：

The keys to push the goods are goods attributes and the record that users browse the commodity. This paper gives the definition of product attributes and the similarity calculation formulation based on the goods interaction （behavior） information, and then an improved slope one recommendation collaborative filtering algorithm is put forward. In the algorithm, the binary string based on the commodity name is proposed to measure the hamming distance based on the commodity name attribute. The effectiveness of the proposed algorithm is verified by the commodity information, customer information, customers for goods browsing records and commodity sparse information, which are provided by an e-commerce site platform. By comparing with the basic slope one algorithm and the weighted slope one algorithm, the rationality of the algorithm is proved.