Dataset of Public Perception of QRIS Usage and Security: A Topic Modeling Analysis of Social Media Narratives
Description
QRIS (Quick Response Code Indonesian Standard) has become a central pillar of Indonesia's digital payment ecosystem, yet public discourse around its security challenges and fraud experiences remains largely absent from academic literature. Earlier studies mostly depended on questionnaire-based methods using TAM and UTAUT frameworks, which only measured risk as subjective pre-use perceptions and restricted users from sharing their genuine post-adoption experiences. This study identifies security gaps, trust gaps, and user-education gaps in the QRIS ecosystem by analyzing organic user comments using the Latent Dirichlet Allocation (LDA) algorithm within the CRISP-DM framework. A total of 1,573 comments were collected from 10 TikTok videos discussing QRIS security and fraud cases; after preprocessing, 1,324 clean comments were retained for analysis. The coherence score analysis confirmed 2 optimal topics. Topic 0, named "QRIS Information Exploration and User Interaction" (26.28%), reflects users actively seeking security information from peers due to the absence of accessible official guidance. Topic 1, named "QRIS User Experience, Concerns, and Payment Practices" (73.72%), directly reveals real fraud experiences including sticker manipulation and phishing, accompanied by a predominance of negative sentiment (50.00%). When mapped against Khando's (2022) five-category digital payment challenge framework, these findings confirm three critical gaps previously undetected by adoption-focused studies. These results provide practical insights and actionable recommendations for payment service providers, Bank Indonesia, and policymakers seeking to strengthen consumer protection within Indonesia's rapidly growing digital payment ecosystem.