摘要:所谓个性化推荐,指的是系统根据网站用户的点击,搜索,购买和评价等行为得出用户的偏好,然后推荐商品给用户。近年来电子商务蓬勃发展,商品数量日益膨胀,顾客在找到自己喜欢的商品之前通常需要花费大量的时间和精力,在大数据膨胀的今天,大量与自身无关的信息的浏览常常使人感到力不从心。本文中通过利用用户在2000年对约4000部电影和书籍的100万个匿名评分信息的数据集和Mahout推荐算法,实现了三种不同方式的推荐:以用户之间相似度为基础的推荐引擎,以内容之间相似度为基础的推荐引擎,以及以Slope One 算法为基础的推荐,为不同的用户推荐满足其兴趣口的电影和书籍,实现个性化推荐。推荐引擎具有良好的发展和应用前景,既能够显著提升企业的商业利益,也能够让用户享受更好的个性化服务。39186 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