基于相似案例分析的电子商务个性化推荐方法研究

发布时间：2018-05-20 18:43

本文选题：电子商务 + 个性化推荐　；参考：《东北大学》2014年硕士论文

【摘要】：电子商务的发展为客户提供了更多的可选商品,同时也带来了“信息超载”问题。客户在面对海量商品时往往显得很迷茫。为此,电子商务个性化推荐应运而生。它利用电子商务网站平台,模拟商店的销售人员向客户推荐商品,克服了信息过载的问题。对于客户来说,电子商务个性化推荐能帮助其从海量商品信息中发现自己感兴趣的商品；对于商家来说,电子商务个性化推荐能帮助其实现差异性经营,同时提高网站的交叉销售,为其带来巨大的经济利益。目前,关于电子商务个性化推荐问题的研究已经引起了企业界及学者界的广泛关注,并取得了一些研究成果。然而,已有的研究成果很少关注客户的分类,以及不同类型客户的偏好分析。此外,商品质量水平和商家在销售过程中提供的服务水平是影响客户做出是否进行购买决策的两个重要因素。因此,需要对电子商务个性化推荐问题进行进一步研究,即按照客户是否给出需求将客户进行分类,并分别给出有针对性的推荐方法,同时在推荐中综合考虑质量和服务水平,是一个值得关注的研究课题,具有重要的理论意义和现实意义。已有研究成果表明,基于相似案例的分析方法符合客户的思维特点,推荐结果容易被客户所接受,因此,将基于相似案例分析的方法引入到电子商务个性化推荐中,并同时考虑客户的分类以及商品的质量水平和商家的服务水平的研究是有必要的。本文对基于相似案例分析的电子商务个性化推荐方法进行了深入研究,主要开展了以下几个方面的研究工作：(1)给出了基于相似案例分析的电子商务个性化推荐问题的描述及研究框架。为了有针对性地分析客户信息,将电子商务个性化推荐问题按照客户在登录网站时是否具有需求分为两类：考虑客户潜在需求的电子商务个性化推荐问题和考虑客户给出需求的电子商务个性化推荐问题。此外,分别给出了基于相似案例分析的电子商务个性化推荐问题的一般性描述以及研究框架。这些基础性的研究工作为电子商务个性化推荐的研究提供了理论指导框架和分析框架,并为研究电子商务个性化推荐的扩展与应用提供了坚实的基础。(2)研究了考虑客户潜在需求的电子商务个性化推荐方法。具体地,针对具有潜在需求的客户,提出了一种基于相似案例分析的考虑客户潜在需求的推荐方法。首先,综合考虑客户的注册信息、行为信息和评价信息对该类型的客户的偏好分析,给出考虑客户潜在需求的案例表示与问题描述；然后,根据客户注册信息计算案例间的相似度,构建了相似客户注册信息案例集；再根据客户是否有购买行为分别计算关于客户行为的案例相似度,并构建了初始商品集；最后,针对初始商品集,综合考虑质量和服务水平,从而确定出推荐商品集。(3)研究了考虑客户给出需求的电子商务个性化推荐方法。具体地,针对给出需求的客户,提出了一种基于相似案例分析的考虑客户给出需求的推荐方法。首先,综合考虑客户的需求信息、行为信息和评价信息对该类型的客户的偏好分析,给出了考虑客户给出需求的案例表示与问题描述；然后,根据客户需求信息计算客户需求与商品集中商品的相似度,构建了需求商品集；再根据客户是否有购买行为分别计算目标案例与购买过需求商品集中商品的案例关于客户行为的相似度,并构建了初始商品集；最后,针对初始商品集,综合考虑质量和服务水平,从而确定出推荐的商品集。本文提出的基于相似案例分析的电子商务个性化推荐方法,考虑了客户类型、质量和服务水平对个性化推荐的影响,为解决电子商务个性化推荐问题提供了可参考的方法。
[Abstract]:The development of e-commerce provides more optional items for customers, and it also brings "information overload". Customers tend to be confused when facing large quantities of goods. For this reason, personalized recommendation of E-commerce arises at the historic moment. It uses e-commerce website platform to simulate store salespeople to recommend goods to customers and overcome the letter. For customers, personalized recommendation of e-commerce can help them find the goods they are interested in from mass commodity information. For businesses, personalized recommendation of e-commerce can help to realize the difference operation, at the same time, improve the cross selling of the website, and bring huge economic benefits to them. The research on personalized recommendation of sub business has attracted wide attention from the business community and the scholars, and has obtained some research results. However, the existing research results seldom pay attention to the classification of customers and the preference analysis of different types of customers. In addition, the quality level of goods and the service level provided by the merchants in the process of sales are the shadow. The customer makes two important factors to decide whether to make a purchase decision. Therefore, it is necessary to further study the personalized recommendation problem of e-commerce, that is, to classify the customers according to the needs of the customers, and to give the targeted methods of recommendation respectively. At the same time, it is worth to consider the quality and service level in the recommendation. The research topic of concern is of great theoretical and practical significance. The existing research results show that the analytical methods based on similar cases conform to the customer's thinking characteristics and the recommended results are easily accepted by the customers. Therefore, the method based on similar case analysis is introduced into the personalized recommendation of the electric sub business, and the customer's score is considered at the same time. It is necessary to study the quality level of the class and the commodity and the service level of the business. This paper makes an in-depth study on the personalized recommendation method of e-commerce based on similar case analysis. The main research work is carried out in the following aspects: (1) the description of personalized recommendation of e-commerce based on similar case analysis is given. In order to analyze the customer information, the problem of e-commerce personalized recommendation is divided into two categories according to the requirements of the customer on the web site: the personalized recommendation of electronic commerce considering the customer's potential demand and the personalized recommendation of the electric sub commerce considering the customer's requirement. The general description and research framework of e-commerce personalized recommendation based on similar case analysis are presented. These basic research provides a theoretical framework and an analytical framework for the research on personalized recommendation of e-commerce, and provides a solid foundation for the study of the extension and application of personalized recommendation in e-commerce. (2) This paper studies the personalized recommendation method of e-commerce in consideration of the potential needs of customers. Specifically, a recommendation method based on similar case analysis is proposed for customers with potential needs. Firstly, the preference of customer's registration information, behavior information and evaluation information is considered. The case representation and problem description that consider the customer's potential demand is given. Then, according to the similarity of the customer registration information, a case set of similar customer registration information is constructed, and the case similarity of customer behavior is calculated according to whether the customer has the purchase behavior, and the initial commodity set is constructed. Finally, the target is set up. Initial commodity set, comprehensive consideration of quality and service level, so as to determine the recommended commodity set. (3) study the personalized recommendation method of e-commerce in consideration of customer requirements. Specifically, a recommendation method based on similar case analysis is proposed for customers to give needs based on similar case analysis. First, consider the customers comprehensively. The demand information, behavior information and evaluation information on this type of customer's preference analysis, give the case representation and problem description considering the customer's requirement, then, calculate the similarity between the customer's demand and the commodity centralized commodity according to the customer's demand information, and build the demand quotient set; and then according to whether the customer has the purchase behavior or not, We calculate the similarity between the target case and the case about the purchase of the commodity in the demand commodity, and build the initial commodity set. Finally, considering the quality and service level of the initial commodity set, the recommended commodity set is determined. The personalized recommendation method based on the phase like analysis in this paper is based on the phase like case analysis. Considering the influence of customer type, quality and service level on personalized recommendation, it provides a reference method for solving personalized recommendation problem in e-commerce.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F724.6

【相似文献】