Text Mining as a Market Intelligence Instrument: A Case Study of Coffee Products on the Tokopedia Platform

Research Aim: This study aims to comprehensively map the landscape of Arabica coffee products on Tokopedia through the integration of three text mining methods: zero-shot classification, word cloud analysis, and text network analysis (TNA). Design/Methods/Approach: A descriptive-exploratory approach based on text mining was applied to 132 listings of Arabica coffee products from Tokopedia (April 2026) using the Social X platform. Three analytical methods were implemented sequentially: zero-shot classification with a multilingual transformer model, word cloud analysis, and TNA with community detection using the Louvain algorithm. Research Finding: Zero-shot classification revealed that 81.06% of products are ready-to-drink packaged coffee. The word cloud identified three main pillars in product naming: varietal terminology (arabica/arabika), geographic indicators (gayo, toraja, aceh), and product form descriptors. TNA resulted in a network of 100 nodes and 2,847 edges with ‘kopi’ as the absolute hub (degree centrality = 1.00) and six thematic communities. Theoretical Contribution/Originality: This study is the first to simultaneously integrate zero-shot classification, word cloud, and TNA to analyze Arabica coffee products in the Indonesian marketplace, providing a multidimensional analytical framework. Practitioner/Policy Implication: The findings serve as a reference for business actors to design effective product naming strategies, for platform managers to develop precise categorization, and for policymakers to understand the dynamics of digital coffee marketing in Indonesia. Research Limitation: The study is limited to 132 listings from a single keyword on Tokopedia over two days and does not yet include consumer reviews or actual sales data.