Mapping Consumer Sentiment and Network Dynamics of Perfume Co-Branding Discourse on X/Twitter: A Computational Social Network Analysis Approach

Abstract

The rapid growth of the global fragrance industry has driven brands to adopt co-branding strategies to strengthen brand equity and expand market reach, yet consumer responses to these collaborations in digital spaces remain fragmented and difficult to predict. This study analyzes the distribution of consumer sentiment, identifies key actors in interaction networks, and explores the extent to which Twitter-based data can complement the evaluation of fragrance co-branding strategy. Using an exploratory social media analytics approach grounded in Digital Public Sphere theory and Social Network Theory, this study integrates Social Network Analysis with six computational modules, wordcloud analysis, sentiment analysis, text network analysis, emotion analysis, trend analysis, and zero-shot classification, applied to 300 public tweets collected via the SocialX platform during 1–12 January 2026 using fragrance and brand collaboration keywords. Results show that public discourse was lexically dominated by neutral sentiment (88%), while zero-shot classification of the same corpus yielded a positive-leaning distribution (82.33%); these are interpreted as two distinct constructs, evaluative polarity versus semantic stance, alongside a Sentiment Index of +0.28 and a dominant happy-emotion classification (92%). Network analysis identified a small number of actors occupying central or bridging positions in information dissemination, and trend analysis detected two major activity peaks on January 1–2, 2026, coinciding with the New Year transition. These findings offer theoretical implications for applying Digital Public Sphere and Social Network Theory jointly to fragrance co-branding discourse, and practical implications for brand managers seeking to monitor how collaboration discourse is expressed and circulates online.