Text Mining and Sentiment Analysis for Mobile Banking Service Quality Measurement: A Cross-Sectional Study of Turkish Private Banks

Abstract

This study proposes a methodological framework for measuring mobile banking service quality through text mining of customer-generated content. The research employs sentiment analysis techniques to evaluate service quality dimensions across three prominent Turkish private banks based on 9,547 Google Play Store reviews. Following a systematic preprocessing protocol, reviews underwent sentiment classification using a Naive Bayes algorithm, achieving 87.3% accuracy. The analysis revealed distinct patterns across five service quality dimen-sions: Practicality emerged as the strongest dimension (mean score: 0.75), while Sociality demonstrated the most significant deficiency (mean score: 0.32). Statistical comparison identified significant inter-bank differences, with İşCep demonstrating superior overall performance (0.67), followed by Garanti BBVA (0.65) and TEB (0.58). Network analysis of keyword co-occurrences illuminated the semantic structure of customer discourse, revealing distinctive terminological communities within dimensional frameworks. The methodology trans-cends traditional survey-based approaches by providing continuous, scalable quality assessment derived from authentic customer expres-sions. Moreover, the integration of dimensional analysis with complaint pattern identification establishes clear priorities for service en-hancement initiatives. This research advances both the theoretical understanding of mobile banking service quality dimensions and the practical implementation of computational text analysis in service quality measurement frameworks.