Digital wallet is one of the financial technology that is currently popularly used by Indonesians as a non-cash transaction tool. The more users of digital wallet applications, the number of reviews, comments, and opinions also increases and varies. User reviews are considered very helpful as well as a forum for information because they can assess certain aspects. This study proposes research related to Aspect-based Sentiment Analysis using Multinomial Naïve Bayes to analyze user sentiment towards an aspect, namely service, cost, and security on digital wallet applications and determine the evaluation of system performance using the Multinomial Naïve Bayes algorithm. The data in this study was taken using scraping techniques with keywords from the Google Play Store platform as many as 500 in each aspect. The results of this study show that the 70:30 data division is better than other data division ratios, namely the 80:20, and 90:10 data division ratios, with performance evaluation using accuracy, precision, recall, and f1-score respectively 0.841, 0.844, 0.841, and 0.841.