基于聚类的推荐算法.docxVIP

曲靖师范学院 本科生毕业论文 论文题目: 一种基于项目聚类的推荐算法 作者、学号：何芸娜 2010112142 学院、年级：数学与信息科学学院2010级 学科、专业：数学 信息与计算科学 指 导 教 师：刘永财 完 成 日 期：2014年5月27日 曲靖师范学院教务处 一种基于项目聚类的推荐算法 摘 要 推荐系统是帮助用户评估他没有发现的内容,从而克服信息超载的一种有效工具.对于推荐系统的研究,既有重大的社会意义,又有重大的经济价值.推荐系统早在上世纪九十年代就已经被提出并进行了广泛的研究. 在现代信息爆炸的年代,用户在网上留下的评分数据成了一个很大的数据库.本文介绍了一种基于项目聚类的协同过滤推荐算法及比较简单实用的聚类分析算法—k-means算法,利用该推荐算法合理开发并利用这些资源.主要通过k-means算法求出根据项目相似性度量,从而对项目进行聚类. 文中介绍的推荐算法,其核心目的在于解决数据稀疏性优势的基础上,使用聚类分析技术对原始信息进行处理,通过简单实用的k-means聚类算法将用户的行为模型转化为兴趣模型从而实现了更精准的推荐. 关键词:推荐系统；聚类分析；相似性度量；k-means算法 Based on the project clustering recommendation algorithm Abstract:Recommendation system is to help users assess the content he did not found an effective tool to overcome the information overload. Recommendation system for the study of both major social significance, but also of great economic value. Recommendation system early in the last century ninety years has been proposed and carried out extensive research. In the modern era of information explosion, leaving the score in the online user data into a large database. This paper describes a project-based clustering collaborative filtering algorithm is relatively simple and practical clustering algorithm-k-means algorithm, using the recommended algorithm development and rational use of these resources. mainly determined by k-means algorithm based on project similarity measure, so the project cluster. This article describes the recommendation algorithm, its core purpose is to solve the data sparsity-based advantages, the use of cluster analysis techniques to process the original information through simple and practical k-means clustering algorithm to model the behavior of the user interest model thus transformed into to achieve a more accurate recommendations. Key word: recommendation system clustering analysis similarity measurement k-means arithmetic 目 录  TOC \o 1-3 \h \z \u  HYPERLINK \l _Toc389164133 1引言  PAGEREF _Toc389164133 \h 1  HYPERLINK \l _Toc389164134 2聚类分析  PAGEREF _Toc389164134 \h