Sentiment Analysis of Chatting Application Reviews on Google Play Store Using Naïve Bayes Classifer Alghoritm

Abstract

Chatting application is a medium used to connect two or more people through social media platforms. Based on the results of the survey report, there are 5 chat applications that are often used as a medium of communication, including WhatsApp, Facebook, Telegram, Instagram and Line applications. This research aims to see the sentiment of chat application users, and see how naive bayes performs in analyzing the sentiment of chat application users.  The purpose of sentiment analysis in this research is to assess whether a comment related to an issue is negative or positive, as well as a guide in improving the quality or service of a product. From the analysis results obtained, the Naïve Bayes model showed mixed performance depending on the type of application and sentiment. The model generally showed better performance in identifying positive reviews, especially on Facebook, Telegram, and Instagram apps, where recall reached 100%. However, the model performed very poorly in identifying neutral reviews across all apps. To increase accuracy and more balanced sentiment detection capabilities, improvements in data preprocessing, handling data imbalance, or the use of more complex classification methods are needed.