个性化推荐系统能很好地解决互联网中信息过载的问题,传统推荐系统存在着商家较为分散、隐私容易泄漏的问题.提出了一种基于中间代理的电子商务智能推荐系统,利用内容过滤技术进行推荐,在考虑用户隐私的基础上使用向量空间模型挖掘用户的兴趣偏好和商品的特征评价,引入时间遗忘函数以处理兴趣变化问题,根据收集的信息产生推荐序列,针对重点难点问题提出了解决方案.采用Movielens数据集进行的实验结果表明,该方法能提供较好的推荐准确度与计算性能.%Personalized recommendation systems can be a good solution for the internet information overload issues. In traditional recommendation systems, more dispersed business and privacy leakage issue can not be solved. Propose an agent-based E-commerce smart recommendation system and utilize content filtering techniques to make recommendation. Based on the privacy of users, use vector space model to mine user preferences and the characteristics evaluation for commodity, introduce the time forgotten function to deal with interest change,generate recommendation sequences based on the collection of information, present the solution for some important and difficult problems. The experiment on data set got from Movielens shows that proposed method can improve the accuracy of the predication and the performance of computing.
展开▼
