大数据视角下非结构化文本数据的顾客满意度研究

发布时间：2018-04-12 21:11

本文选题：网络购物 + 情感分析　；参考：《首都经济贸易大学》2017年硕士论文

【摘要】：在这个互联网高速发展的时代,电子商务企业竞争激烈,顾客对产品的满意程度(Customer Satisfaction)是关系到企业存亡的问题,更会影响一个国家和地区电子商务的可持续发展。开展消费者满意度研究,能够及时了解和掌握消费者对品牌的认知度、忠诚度和满意度情况,为相关行业企业提升品牌价值提供参考意见,对于推动国家实施品牌战略,培育一批具有国际影响力的品牌具有重大意义。本论文主要利用大数据分析的方法,基于非结构化文本数据研究顾客满意度。具体研究内容包括以下方面:首先在充分研究国内外满意度模型以及结构方程模型理论的基础上,根据顾客满意度的影响因素,初步设定了适合本研究的CSI测评模型;其次,通过Python网络爬虫技术,抓取了4大购物网站(天猫、京东、国美在线、苏宁易购)中7个电器数码类品牌产品的消费者评论数据;然后,对商品评论利用结巴分词的方法进行分词,结合评论语义,构建了消费者满意度评论词库,从评论词中提炼出影响消费者对品牌满意度的实体词汇并结合修饰词进行量化,实现了文本数据的结构化;最后,将结构化后的文本数据应用于之前预设的CSI测评模型,并对模型加以改进,得到了适用于文本数据的CSI测评模型,通过对模型参数的估计,测算了顾客满意度,并利用绩效分析方法,按照隐变量——显变量——实体的思路,总结出了具体到实体层面的品牌满意度提升途径。本论文主要创新之处有两点,一是在现有的CSI框架下,将传统的、以问卷调查为主要手段的主观数据收集手段变为收集客观存在的、海量的消费者对商品评论数据,能够更为客观、真实地反映消费者对所购商品的看法;二是把大数据视角下的文本挖掘引入到了传统的满意度测度模型中,大数据分析手段与传统CSI模型结合起来,为满意度模型的研究提供新的思路。通过研究,本论文认为品牌提升的关键需要从感知质量、期望质量和品牌形象三方面入手,具体到实体层面,则需从外观、手感、外壳、通话、电池等产品质量方面,物流、客服等服务方面,促销、折扣、返券等优惠方面以及口碑、宣传、诚信等品牌形象方面加以提高。
[Abstract]:In this era of rapid development of the Internet, the competition of e-commerce enterprises is fierce, and the degree of customer satisfaction is related to the survival of enterprises, and will also affect the sustainable development of e-commerce in a country and region.Carry out the research of consumer satisfaction, can understand and grasp the consumer's recognition, loyalty and satisfaction of brand in time, and provide reference for enterprises in related industries to promote brand value, and promote the implementation of brand strategy for the country.It is of great significance to cultivate a group of brands with international influence.This paper mainly uses big data analysis method to study customer satisfaction based on unstructured text data.The specific research contents include the following aspects: firstly, on the basis of the domestic and foreign satisfaction model and structural equation model theory, according to the influencing factors of customer satisfaction, the CSI evaluation model suitable for this study is preliminarily established; secondly,Through Python web crawler technology, this paper grabs the consumer comment data of 7 electronic digital brand products from four major shopping websites (Tmall, JingDong, Gome online, Tmall easy to buy).In this paper, we use the method of stutter participle to segment commodity reviews, combining with the comment semantics, construct the consumer satisfaction comment lexicon, extract the entity words that affect the consumer satisfaction to the brand from the comment words, and combine the modifiers to quantify them.Finally, the structured text data is applied to the pre-default CSI evaluation model, and the model is improved to obtain the CSI evaluation model suitable for text data.This paper calculates customer satisfaction, and summarizes the way to enhance brand satisfaction at the entity level according to the idea of implicit variable, explicit variable and entity by using the method of performance analysis.There are two main innovations in this paper. One is that under the existing CSI framework, the traditional subjective data collection method, which mainly uses questionnaire as the main means, is changed into the collection of objectively existing, massive consumer comment data on commodities.It can more objectively and truly reflect the consumer's view of the goods purchased. Secondly, the text mining from big data's perspective is introduced into the traditional satisfaction measurement model, and big data analysis method is combined with the traditional CSI model.It provides a new idea for the research of satisfaction model.Through the research, this paper thinks that the key to brand promotion needs to start from three aspects of perceived quality, expectation quality and brand image, and to the physical level, it needs from the aspects of appearance, handle, shell, telephone, battery and other product quality, logistics, etc.Customer service, sales promotion, discounts, coupons and other concessions as well as word of mouth, publicity, integrity and other brand image to be improved.
【学位授予单位】：首都经济贸易大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F274

【参考文献】