丝绸2025，Vol.62Issue(5)：22-33,12.DOI:10.3969/j.issn.1001-7003.2025.05.003
基于社会化聆听的服装品牌个性维度构建与量化
Construction and quantification of clothing brand personality dimensions based on social listening
摘要
Abstract
Brand personality,as a core element of consumer-brand emotional connection,is a crucial component of brand equity.It holds significant symbolic value in fashion consumption and directly influences consumers' brand choices and loyalty.While Aaker's five-dimensional framework of brand personality is widely applied,there are still certain limitations in current research:first,there is an over-reliance on official brand communication materials,neglecting the role of consumers as active agents in shaping brand personality;second,there has been insufficient integration of the vast amounts of user-generated content in the social media era.Extracting brand personality information from consumer expressions on social media is not only feasible but also necessary.Given these issues,this paper aims to expand the existing research paradigm of brand personality from the perspective of social media.The primary goal of this study is to develop a framework for brand personality dimensions in the fashion industry based on social listening data and natural language processing,and to implement a quantitative analysis of brand personality. This study consists of five stages:data collection,original lexicon construction,brand personality lexicon selection,brand personality dimension construction,and brand personality measurement.In the data collection and preprocessing stage,the study focuses on the characteristics of social media contexts,selecting the Xiaohongshu platform,which has a high level of user engagement in fashion content sharing,as the data source.Based on the"Global 500 Brands"list,28 global fashion brands such as Gucci and ZARA were selected.Relevant content was retrieved using keyword combinations of "brand name+runway/style",and a corpus containing 1.46 million words was built.In the data cleaning and word segmentation process,the study utilized the THULAC tool to segment the original text,retaining only nouns,adjectives,and verb phrases,while removing stopwords and low-frequency words.Additionally,negative words were filtered using Dalian University of Technology's sentiment lexicon,resulting in an original brand lexicon consisting of 972 words.In the brand personality lexicon selection phase,the study first employed the text-embedding-ada-002 word embedding model to map the original vocabulary into a high-dimensional semantic space.Then,based on Keller's brand knowledge structure and the characteristics of fashion products,zero-shot learning classification labels were constructed,and the similarity between the words and each label was calculated.As a result,176 symbolic words highly correlated with brand personality were selected,forming the brand personality lexicon.In the brand personality dimension construction stage,a combined approach of data-driven analysis and theoretical validation was used.First,clustering analysis determined that the optimal number of clusters for the word vectors was 28 sub-dimensions.Then,three brand management experts were invited to validate the clustering results to ensure the rationality of the dimension construction.Following the mainstream paradigm of brand personality dimension construction,the study employed the Big Five Personality Model from trait theory,categorizing the 28 sub-dimensions into five main dimensions,thus forming a complete brand personality dimension system for fashion brands.Finally,in the brand personality measurement phase,the study quantified the weight of each dimension and word in the brand personality lexicon using the TF-IDF algorithm,enabling quantitative and visual analysis of brand personality. The innovation of this study is reflected in three aspects.First,in terms of methodology,this study proposes and constructs a systematic research framework for fashion brand personality,addressing the traditional limitation of insufficient focus on consumer self-expression in shaping brand personality,thus enriching both the theoretical and practical perspectives of brand personality research.Second,in terms of theoretical contribution,the research identifies emerging sub-dimensions such as "casual" (agreeableness dimension)and "understated" (emotional stability dimension) in consumer language expressions on social media,which more accurately capture contemporary consumers' brand personality expressions,expanding the applicability of brand personality theory in the social media context.Third,in terms of practical application,this study introduces a new set of brand personality dimensions suitable for the social media age and develops a TF-IDF-based quantification method for brand personality,which can be used for real-time monitoring of brand personality recognition and provide data-driven decision support for brand management. The findings of this study indicate that social media data can effectively reveal consumer perceptions of brand personality.The study successfully constructed a brand personality dimension framework based on the Big Five Personality Model and Keller's brand knowledge structure,encompassing five main dimensions,28 sub-dimensions,and 176 core brand personality words,providing a quantitative tool for fashion brand personality research.Future research could expand the brand sample range to include more small and medium-sized brands and integrate data from multiple platforms to enhance the external validity of the study.Additionally,a real-time brand personality monitoring system could be further developed to provide more precise data support for brand strategy formulation.关键词
品牌个性/社会化聆听/自然语言处理/情感分析/词嵌入/GPT模型/零样本学习/层次聚类Key words
brand personality/social listening/natural language processing/sentiment analysis/word embedding/GPT model/zero-shot learning/hierarchical clustering分类
轻工业引用本文复制引用
章冰珏,宋琨,顾新荣..基于社会化聆听的服装品牌个性维度构建与量化[J].丝绸,2025,62(5):22-33,12.基金项目
中央高校基本科研业务费专项基金项目(2232020G-08) （2232020G-08）
