Implementation of Classification Algorithm for Sentiment Analysis: Measuring App User Satisfaction

Abstract

Google Play Store is the official app store for Android devices from Google that offers rating and review features. This feature on the platform is a source of data for sentiment analysis in research on app user satisfaction. The purpose of this study is to provide an overview of app user satisfaction and evaluate the accuracy of the algorithms used. The algorithms compared include Support Vector Machine (SVM), namely linear, rbf, sigmoid, and polynomial kernels with Naïve Bayes Classifier (NBC). The key variables analyzed include perceived usefulness, perceived ease of use, relia-bility, responsiveness, and website design. The results showed that the SVM algorithm with a linear kernel achieved the highest accuracy of 95.23% compared to the NBC algorithm of 91.43%. For other accuracy results, rbf kernel 94.35%, sigmoid kernel 95.19% and polynomial kernel 93.31%. In addition, the results of sentiment analysis on application user satis-faction revealed that 75% of users were dissatisfied, with the service indicator having the highest number of negative sen-timents. These findings suggest that sentiment analysis can be an effective tool for companies to measure and improve user satisfaction. In addition, these results can also be a useful reference for new users in assessing apps before using them.

Keywords

• google play store
• naïve bayes
• sentiment analysis
• support vector machine
• user satisfaction