摘要
协同过滤技术是目前运用最广泛的个性化推荐技术之一,但随着系统规模的不断扩大,用户评分数据极端稀疏等问题使其推荐质量严重下降。因此,文章提出将维数简化和聚类的方法运用到协同过滤技术中,从而较好地解决协同过滤推荐技术中存在的稀疏性、扩展性等问题,快速准确地产生个性化推荐结果。
Collaborative filtering is one of the most widely used technologies in personality recommendation. However, the efficiency of this technology declines by the increasing number of users and items, which results in extremely sparse data of users' assessments and other problems. In the paper, the methods of dimensionality reduction and clustering are proposed,which may solve the problems of sparsity and scalability, so that accurate results of personality recommendation can be obtained quickly.
出处
《合肥工业大学学报(自然科学版)》
CAS
CSCD
北大核心
2008年第7期1059-1062,1148,共5页
Journal of Hefei University of Technology:Natural Science
基金
安徽省自然科学基金资助项目(070412054)
合肥工业大学科学研究发展基金资助项目(062101f)
关键词
个性化推荐
协同过滤
维数简化
聚类
personality recommendation
collaborative filtering
dimensionality reduction
clustering