摘要
针对电子商务推荐系统传统项目相似性计算不足和传统的TOP-N推荐输出缺乏个性等问题,提出了一种改进算法.该算法采用动态生成权重因子的方法计算项目相似性,利用模糊聚类对项目分类,在预测评分公式中引入用户权重因子,并采用集团序方法进行推荐输出以确定候选的N值.实验结果表明,该算法不仅解决了数据稀疏性问题,提高了系统的推荐质量,且推荐结果更个性化.
Aiming at the problem that traditional item similarity of e-commerce recommedation system has shortage in calculation and the recommedation output of TOP-N lacks personalization,an improvement algorithm was put forword.This algorithm calculates item similarity using dynamic weigh factor method,classifies items using fuzzy clustering,introduces user weigh factor in forecasting evaluation formula and carries out recommendation output to confirm candidate N value.The experimental results showed that this algorithm not only solved the problem of sparse data and improved the recommendation quality and enhanced the recommendation personalized degree.
出处
《郑州轻工业学院学报（自然科学版）》
CAS
2012年第2期80-83,共4页
Journal of Zhengzhou University of Light Industry:Natural Science
关键词
协同过滤推荐算法
模糊聚类
集团序
项目相似性计算
collaborative filtering recommendation algorithm
fuzzy clustering
aggregative rank
item similarity calculation
