基于个性化推荐引擎组合的推荐系统的设计与实现

发布时间：2018-06-24 08:49

本文选题：推荐引擎组合 + 个性化推荐　；参考：《华南理工大学》2012年硕士论文

【摘要】：近年来，随着互联网的高速发展，现已进入到数据过载阶段，在构建互联网网站时需要考虑如何方便用户在庞大的数据中获取他所需的信息。传统的互联网企业的解决方式有两种，第一种是信息目录，如雅虎和新浪；第二种解决方式是搜索引擎，如谷歌和百度。两者的共同点是用户对自己的需求非常明确。但是互联网用户的现状是：大部分互联网用户对需求不明确。因此，推荐引擎应运而生。它主动向用户推荐用户可能喜欢的物品，不需要用户主动提供任何输入，而是通过在后台记录并分析用户行为数据，最后把分析结果作为个性化推荐推送给用户。本论文首先介绍了课题的研究背景，国内外研究现状及水平等，阐述了系统相关的理论基础，探讨了推荐系统相关技术与工具，，确定了系统的体系架构。其次，详细介绍了推荐系统的架构、以及相关推荐系统的具体实现过程。最后利用Mahout框架把推荐引擎应用在大规模数据环境中。本论文的主要贡献包括：（1）分别实现了三种推荐引擎，并通过组合各种推荐策略的优点生成推荐列表。（2）记录并分析用户的反馈信息，以调整各种推荐引擎组合方式；以个性化的推荐引擎组合推荐物品；实验表明，使用了个性化引擎组合的方式产生的推荐结果比使用单个推荐引擎总体上获得更好的效果。（3）利用Hadoop的Mahout框架，把该推荐系统应用到大规模数据环境中。
[Abstract]:In recent years, with the rapid development of the Internet, it has entered the stage of data overload. When constructing the Internet website, it is necessary to consider how to facilitate users to obtain the information they need in the huge data. There are two traditional solutions for Internet companies, the first is information directories, such as Yahoo and Sina, and the second is search engines, such as Google and Baidu. The common denominator between the two is that users are very clear about their needs. But the current situation of Internet users is that most Internet users are not clear about their needs. Therefore, the recommendation engine came into being. It actively recommends the items that the user may like, does not need the user to provide any input voluntarily, but records and analyzes the user behavior data in the background, finally pushes the analysis result as the personalized recommendation to the user. This paper first introduces the research background, the current situation and the level of the research at home and abroad, expounds the theoretical basis of the system, discusses the related technologies and tools of the recommendation system, and determines the architecture of the system. Secondly, the architecture of recommendation system and the implementation process of related recommendation system are introduced in detail. Finally, the recommendation engine is applied to large scale data environment using Mahout framework. The main contributions of this paper are as follows: (1) three recommendation engines are implemented, and a recommendation list is generated by combining the advantages of various recommendation strategies. (2) users' feedback information is recorded and analyzed. In order to adjust the combination of various recommendation engines; to combine the recommended items with the personalized recommendation engine; the experimental results show that, The recommendation results obtained by using the combination of personalized engines are better than that of single recommendation engines. (3) using the Mahout framework of Hadoop, the recommendation system is applied to the large-scale data environment.
【学位授予单位】：华南理工大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP311.52

【参考文献】