基于大语言模型的消费者社交媒体表征分析:以淘宝退货服务为例  

Analysis of Consumer Social Media Representation Based on Large Language Models:A Case Study of Taobao’s Return Service

作　　者：马少辉 王炜辰 MA Shao-hui;WANG Wei-chen(Business School,Nanjing Audit University)

机构地区：[1]南京审计大学商学院

出　　处：《消费经济》2024年第6期85-98,共14页Consumer Economics

基　　金：国家自然科学基金面上项目(72072092);江苏省研究生科研与实践创新计划项目(SJCX24_1186)。

摘　　要：在数字化时代,企业管理者需要通过大数据技术积极倾听消费者对产品和服务的反馈,以不断改进其运营绩效。社交媒体是消费者购物后自由分享和讨论他们对产品和服务感知的重要渠道。对海量消费者不断发出的杂乱文本信息进行分析、总结和提炼,以得到消费者群体的核心表达是一个困难的问题。文章提出了基于大语言模型和社会表征理论进行社交媒体分析的新方法框架,并以新浪微博推文数据源的淘宝退货服务为例,说明该方法框架的具体应用。案例分析发现,退货困难、产品质量、物流配送、包装问题等是消费者反映的核心问题。其中退货困难、产品质量问题往往伴随消费者更负面的情绪。与传统主题分析技术相比,新方法框架可以帮助企业从消费者社交媒体数据中提炼多维度、多层次的社会表征,为了解消费者对企业产品和服务的感知提供更准确、更深入、更全面的洞察。In the digital age,business managers need to actively listen to consumer feedback on products and services through big data technology in order to continuously improve their operational performance.Social media serves as an important channel for consumers to freely share and discuss their perceptions of products and services after shopping.Analyzing,summarizing,and refining the vast amount of messy text information continuously generated by consumers to obtain the core expressions of consumer groups is a challenging task. This studyproposes an analytical framework for social media analysis based on large language models andsocial representation theory, and uses Sina Weibo tweets to analyze Taobao return services as anexample to illustrate the specific application of this framework. Case analysis reveals that returndifficulties, product quality, logistics distribution, packaging issues, etc., are core issues reflectedby consumers. Among them, return difficulties and product quality issues often accompany morenegative emotions from consumers. Compared with traditional topic analysis techniques, the newanalytical framework can help enterprises refine multi-dimensional and multi-level socialrepresentations from consumer social media data, obtain more accurate, in-depth, andcomprehensive insights into consumer perceptions on their products and services.

关 键 词：社交媒体分析 大语言模型 社会表征 情绪分析 退货服务 

分 类 号：F713.36[经济管理—产业经济]