基于Mahout的推荐系统的设计与实现
毕业论文关键词：Mahout；个性化推荐系统；协同过滤
Design and Implementation of the Recommender System based on the Mahout
Abstract：The personalized recommender system refers to that the system according to the website user click, search, purchase and evaluation to obtain the user's preferences to recommend. E-commerce sites recommend the user the goods in which he may be interested to improve the user experience. In recent years, scale of e-commerce booming, the number and species growing, customers often need to spend a lot of time and energy before finding their favorite goods. In the expansion of the large data, the browsing of information which has nothing to do with customers often makes them feel overwhelmed. This article makes the use of data sets based on the user in 2000 about 1 million anonymous rating of 4000 movies and books and the Mahout algorithm to implement the recommender engines in three different ways: recommendation engine based on user’s similarity, recommendation engine based on item’s similarity and recommendation engine based on the Slope One algorithm. The recommendation engines recommends information to different users to satisfy their interest in the choice of movies and books. Personalized recommender system has a good prospect of development and application。Not only can significantly personalized recommender system improve enterprise business interests, but also can let users enjoy better personalized service.
Keywords：Mahout; Personalized recommender system; Collaborative filtering
目 录
第1章 绪论 1
1.1个性化推荐的背景 1
1.1.1 个性化推荐的原因 1
1.1.2 个性化推荐的优势 1
1.1.3 个性化推荐的前景 1
1.2 推荐算法MAHOUT 1
1.2.1 机器学习 1
1.2.2 介绍MAHOUT 2
1.3 本文研究内容及主要贡献 2
第2章 推荐系统分析与设计 3
2.1 需求分析 3
2.2 推荐引擎选择 3
2.3 协同过滤算法选择--TASTE 4
2.4 推荐引擎的结构 4
2.5 环境配置设计 5
2.6 本章小结 7
第3章 核心推荐算法 8
3.1 推荐算法 8
3.1.1 按数据源划分 8
3.1.2 按模型划分 8
3.2 协同过滤的基本思想 8
3.2.1 用户爱好的采集 8
3.2.2 相似度 9
3.2.3 距离的计算 9
3.2.4 探究用户邻域 9
3.2.5 基于用户的协同过滤算法UserCF 10
3.2.6 基于物品的协同过滤算法ItemCF 10