基于数据挖掘的网上商城个性化推荐模型研究

发布时间：2018-03-02 14:57

本文选题：数据挖掘　切入点：网上商城　出处：《重庆工商大学》2015年硕士论文　论文类型：学位论文

【摘要】：随着信息技术的发展和计算机网络的普及,以高效、便捷、低成本为特点的电子商务得到了迅猛发展。电子商务网站的竞争变得很激烈,伴随着商品数量与日俱增,以用户为中心的时代,在用户选购商品时,需求能否尽快得到满足,决定着电子商务网站能否在竞争中立于不败之地。网上商城个性化推荐变得越来越重要,基于此,本文进行了相关研究。本论文主要关注数据挖掘中应用聚类、决策树和关联规则算法在网上商城推荐模型中的应用。首先介绍数据挖掘的基本理论及相关算法,针对本文采用的三种算法进行了详细的描述。从实际出发,以销售图书的网上商城数据集为依据,探讨了数据挖掘中的数据集预处理过程。在数据整理完善后,使用SPSS Clementine等工具,结合聚类、关联、决策树等算法进行4个个性化推荐模型的构建。首先根据访问习惯和关键词查询情况建立了购买行为预测模型,用于找到有价值的用户。然后根据用户访问网站的网页内容情况,对用户聚类细分找到最常购买的产品,建立产品推荐模型并用C5.0算法对模型进行评估。接下来根据用户首先访问的3个页面和访问页面之间的时间间隔对用户聚类,并找出可能访问的第4个页面进行相应的页面推荐。最后对用户的基本信息特征与商品风格之间建立关联模型,找出用户的特征和购买商品风格之间的关联,对新访客中具有相似特征的用户进行相应风格商品的推荐。针对建立的4个模型都进行了评估与发布,进一步展现推荐结果,然后描述了个性化推荐系统的总体设计以及在应用过程中可能面临的问题,最后指出论文的不足之处并提出一些改进方法。总体来看,通过数据挖掘理论和电子商务实际项目结合,实现网上商城个性化推荐模型的构建,对解决网上商城个性化推荐方面的问题具有重要意义。针对数据挖掘与推荐系统,目前的研究主要集中在推荐算法的改进方面,怎样构建模型以及形成个性化推荐引擎嵌入到电子商务网站的研究较少,本文进行了相应的介绍,为以后这方面的的研究提供了研究思路。
[Abstract]:With the development of information technology and the popularization of computer network, E-commerce, characterized by high efficiency, convenience and low cost, has been developed rapidly. In the era of taking the user as the center, whether or not the demand can be satisfied as soon as possible, determines whether the e-commerce website can be in an invincible position in the competition. This paper mainly focuses on the application of clustering, decision tree and association rules algorithm in online shopping mall recommendation model. Firstly, the basic theory and related algorithms of data mining are introduced. In this paper, the three algorithms used in this paper are described in detail. Based on the data set of online shopping mall which sells books, the preprocessing process of data set in data mining is discussed. Using SPSS Clementine and other tools, combining clustering, association, decision tree and other algorithms to construct four personalized recommendation models. Firstly, according to the visiting habit and keyword query, the purchase behavior prediction model is established. It is used to find valuable users. Then according to the content of the web page visited by the user, the most frequently purchased products are found by clustering the users. The product recommendation model is established and evaluated by C5.0 algorithm. Then the users are clustered according to the time interval between the three pages visited by the user first and the time interval between the access pages. Finally, the correlation model between the user's basic information characteristics and the product style is established to find out the relationship between the user's characteristics and the purchase style. For the new visitors with similar characteristics of the users of the corresponding style of goods recommended. For the establishment of the four models are evaluated and published, to further show the results of the recommendation, Then describes the overall design of the personalized recommendation system and the possible problems in the application process, finally points out the shortcomings of the paper and puts forward some improvement methods. By combining the theory of data mining with the practical project of electronic commerce, the construction of personalized recommendation model of online shopping mall is realized, which is of great significance to solve the problem of personalized recommendation of online mall. The current research mainly focuses on the improvement of recommendation algorithm, how to build a model and how to form a personalized recommendation engine embedded into e-commerce website is less, this paper gives a corresponding introduction. For the future research in this area to provide research ideas.
【学位授予单位】：重庆工商大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F724.6

【参考文献】