A Data-Driven Analysis of Understanding Technostress in Indonesian E-Wallets Using BERTopic

The rapid development of digital wallet services (e-wallets) in Indonesia has revolutionized people's financial transaction patterns. However, the adoption of this technology also has a negative impact in the form of technostress, which is psychological pressure due to interaction with technology that is not optimal. This study aims to identify and categorize the main topics that reflect the form of technostress based on user reviews of the XYZ application on the Google Play Store. The topic mapping process uses 87,030 automatically collected user reviews, then analyzed using a transformer-based topic modeling approach, namely BERTopic. This approach combines embedding from BERT, document clustering using HDBSCAN, and topic representation through class-based TF-IDF (c-TF-IDF), resulting in more semantic and accurate topic mapping than conventional methods. The analysis revealed four dominant factors that trigger technostress, namely: (1) Application & Tool, (2) Entry Point, (3) Service Platform, and (4) Data Protection. Users often report disruptions such as login failures, system errors, transaction errors, and concerns about data security. While some reviews indicate satisfaction with the ease of service, findings show that negative experiences have a significant impact on user trust and loyalty. With a topic coherence score of 0.9986, BERTopic proved effective in generating meaningful clusters. This research contributes to the development of data-driven strategies to improve e-wallet service quality through system improvement, security, and customer service responsiveness.