摘要
个性化推荐服务系统是根据用户历史记录和推荐算法为用户提供其感兴趣的个性化信息或商品的一种自动化工具。针对目前常用的基于协同过滤的推荐算法和基于内容的推荐算法各自存在的问题,本文提出一种结合协同过滤和隐语义分析的混合推荐算法——交替奇异值分解算法ASVD,通过奇异值分解算法对基于项目内容的项目-关键词矩阵和对用户评分信息得到的用户—项目矩阵进行分解过程产生的项目—隐主题矩阵合并优化来消除噪音提高推荐的精确度。实践结果表明,新的混合算法ASVD提高了推荐结果的准确性。
Personalized recommendation service system is based on user history record and recommendation algorithm to provide personalized information or commodities for different users which they may be interested in. According to the problems exist in collaborative filtering recommendation algorithm and content-based recommendation algorithm respectively, this paper presents a recommendation algorithm called ASVD that alternately merge and optimized project-latent topic matrix which is from the process of using singular value decomposition algorithm to decompose the project-keywords matrix based on project content and user rat- ings users-project matrix to eliminate the noise to improve the accuracy of the recommendation. Experiments show that the new hy- brid algorithm ASVD can significantly improve recommendation accuracy.
出处
《计算机与现代化》
2013年第8期64-67,共4页
Computer and Modernization
关键词
协同过滤
内容过滤
个性化推荐
混合推荐
collaborative filtering
content filtering
personalized recommendation
hybrid recommendation
