基于特征事务元素的用户事务聚类方法研究

针对现有用户访问事务聚类算法复杂度较高的问题，文章提出了一种新的用户事务聚类算法。首先对用户事务中访问过的页面浏览次数和时间进行综合统计，并定义了函数，将其映射为页面访问时间属性值；其次通过抽取事务的特征事务元素集，建立新的聚类对象集事务，实现对原有事务的降维。最后给出计算事务相似度的方法．并得到一个模糊相似度矩阵以实现事务聚类。该算法具有较高的准确性，计算复杂度低，扩展性好，可以广泛地应用于电子商务领域。

Studying of a Clustering Method for User Session Based on Feature Elements Set

A new fuzzy clustering algorithm of web session is presented based on the analysis of the drawbacks of traditional methods. Firstly, combine both the user's viewing times and viewing time for a page and maps it into a property value of web access. Then, a User Sessions are descended by extracting the feature elements from a original user session. Finally, the similar user sessions group by fuzzy algorithm. This algorithm is proved to have better accuracy, fewer CPU time and better scalability than others by the experiments. 1.