Mining User Sentiments in P2P Lending: A BERTopic Approach to Understanding User Technostress in Indonesia

Peer-to-peer (P2P) lending has emerged as a popular financial technology (fintech) solution, offering convenient and accessible digital credit services. However, despite its rapid adoption, users often experience challenges related to system functionality, accessibility, and security, which can lead to technostress. In one P2P Lending application, ABC, there have been several complaints related to convenience, user satisfaction in using the application, and a decrease in the number of application transactions. This study aims to identify the main problems of ABC app users by analyzing Google Play Store review data using BERTopic, a transformer-based topic modeling technique. BERTopic is supported by document embedding, document clustering, and topic representation processes. The results show two dominant categories of user concerns: functionality and availability. With an average coherence score of 0.99, BERTopic effectively identifies key pain points and user expectations. The findings provide valuable recommendations for app developers to improve service reliability, enhance security measures, and streamline user experience. Addressing these issues can help reduce technological stress, foster trust, and encourage higher engagement in digital lending services.