联系我们

论文查询结果	返回搜索

论文编号：	14737
作者编号：	2120223638
上传时间：	2024/6/7 11:44:18
中文题目：	电商导购平台中机器生成内容对消费者购买决策的影响
英文题目：	The influence of machine generated content in e-commerce shopping guide platform on consumers'' purchasing decisions
指导老师：	李凯
中文关键字：	机器生成内容；消费者感知价值；消费者购买决策；分类评定模型
英文关键字：	Machine generated content; Consumer purchase intention; Consumer perceived value; Logit model
中文摘要：	电子商务在日趋激烈的竞争中应用的技术手段不断增多，平台分类不断细化，其中的电商导购平台没有购物功能，只抓取各购物平台的商品信息来生成导购内容和链接。电商导购平台的导购内容可以按照生成主体分为机器生成内容（MGC）和用户生成内容（UGC）2种，两种主体生成的文本内容有区别，这在电商导购平台中会影响消费者感知价值，进而影响对消费者购买决策，两种主体在链接平台、爆料商品、不同条件下表现可能不同，这是本文的主要研究内容。 本研究收集了“什么值得买”平台共计5万条数据，数据类别包括文章标题、标签，原文等文本内容；也有商品价格、文章收藏量、评论量、浏览量、点击量和直接订单量等连续数据，每一条数据都标注了用户类型、商品种类以及链接平台。其中由机器创建的文章有14576条，用户创建的文章有33750条，筛选出47267条数据用于分析。本研究使用Python对内容进行文本分析和情感分析，使用Stata和SPSS分别进行实证分析和多元统计分析。 本研究首先通过主效应分析电商导购平台中机器生成内容和用户生成内容在促进消费者浏览量、点击量和订单量指标上的表现差异，发现机器生成内容对消费者购买决策的表现显著不如用户生成内容；其次通过中介效应分析，发现内容质量特征和情感特征是造成这种差异的原因；最后通过异质性分析探究两种主体在导向不同平台和导购不同商品种类上的表现差异，发现存在异质性。在接下来的拓展研究中通过分组回归进一步探究造成平台差异和商品差异的影响因素，发现在机器生成内容的条件下，平台成立时间、用户画像和平台流量是不同平台表现差异的影响因素；实体商品还是虚拟商品是不同商品表现差异的影响因素，最后使用多元统计的主成分分析法对平台和商品表现进行评价，为机器生成内容选择平台和商品提供依据。 本研究发现机器生成内容可以通过提升生成内容特征的丰富性、长效性和互动性来提高消费者感知价值，从而促进消费者购买决策；导购平台使用机器时优先选择流量大、用户画像一致和成立时间短的平台和虚拟商品生成导购内容更有利于消费者做出决策。
英文摘要：	In the increasingly fierce competition, the application of technical means of e-commerce is increasing, and the classification of platforms is constantly refined. The e-commerce shopping guide platform has no shopping function, and only grabs the commodity information of each shopping platform to generate shopping guide content and links. The shopping guide content of the e-commerce shopping guide platform can be divided into machine-generated content and user-generated content according to the generation subject, and the content focus generated by the two subjects has certain differences. On the e-commerce shopping guide platform, the content of this text may have a greater impact on consumers' purchasing decisions, while the price, the number of reviews, the type of goods, and the link platform may have an impact, but this impact is not clear, which is the main research content of this study. This study collected a total of 50,000 pieces of data, including article title, label, original text and other text content; There are also continuous data such as product prices, article favorites, comments, views, clicks, and direct orders, each of which is labeled with user type, product type, and link platform. Among them, 14,576 articles were created by machines, accounting for 29.1%; There were 33,750 articles created by users, accounting for 70.9%, and 47,267 pieces of data were screened for analysis. In this study, Python was used for text analysis and sentiment analysis, Stata and SPSS were used for empirical analysis and multivariate statistical analysis, respectively. This study first analyzed the differences in the performance of machine-generated content and user-generated content in promoting consumer page views, clicks and orders in the e-commerce shopping guide platform, and found that the performance of machine-generated content on consumer purchase decision is significantly worse than that of user-generated content. Secondly, through the mediation effect analysis, it is found that the content quality characteristics and emotional characteristics are the reasons for the difference. Finally, heterogeneity analysis was conducted to explore the performance differences between the two subjects in guiding different platforms and guiding different types of goods, and it was found that machine-generated content in guiding Pindoduo platforms and guiding financial services products had a significantly better impact on consumers' purchasing decisions than user-generated content, with heterogeneity. Since the reasons for such heterogeneity are not clear, in the following expansion research, we further explore the factors that affect the platform differences and commodity differences through grouping regression. It is found that under the condition of machine generated content, platform establishment time, platform user portrait and platform traffic are the factors that affect the performance differences of different platforms. Physical goods or virtual goods are the factors that affect the performance difference of different commodities. Finally, multivariate statistical principal component analysis is used to construct the evaluation system of the platform, which provides the basis for the machine-generated content selection platform and commodities. This research find that machine-generated content can promote consumers' purchase decision by improving the richness, long-term performance and interactivity of generated content features. When using the machine, the platform preferentially selects platforms and virtual goods with large traffic, consistent user profiles and short establishment time to generate shopping guide content to help consumers to make decisions better.
查看全文：	预览  下载(下载需要进行登录)