摘要
传统推荐算法仅依据网络消费者已有偏好信息提供推荐服务,忽略其当前购物状态信息和可能的偏好变化信息。针对这一缺陷,通过分析网络消费者偏好变化特征,提出基于网络消费者偏好预测的推荐算法。该算法综合考虑网络消费者已有偏好信息及其前购物操作行为评估其对商品的偏好,并结合协同过滤思想为其提供有针对性的推荐服务。实验结果表明,基于网络消费者偏好预测的推荐算法能够较好地预测其购物过程中的偏好倾向,显著提高推荐质量和精度。
Traditional recommendation algorithms provide recommendation service based on web consumers＇ previous information without consideration of current shopping information or possibly varied preference information. In this paper, a recommendation algorithm is put forward based on web consumers＇ preference prediction according to variation characteristics of web consumer preference. In order to provide specified service for web consumers, consumers＇ previous preference information and current shopping information is combined to be used to propose a recommendation algorithm integration with collaborative filtering idea. The experimental result indicates that the proposed algorithm can predict consumers＇ current shopping preference much more accurately, while the quality and accuracy of recommendation is conspicuously improved.
出处
《图书情报工作》
CSSCI
北大核心
2012年第4期120-125,共6页
Library and Information Service
基金
国家自然科学基金项目"网络消费者偏好与品牌选择模型研究"(项目编号:70862001)研究成果之一
关键词
电子商务
推荐算法
偏好预测
协同过滤
e-commerce
recommendation algorithm
preference prediction
collaborative filtering
