Deep Learning-based Sentiment Analysis of Public Comments on Military Education Using RoBERTa Algorithm and Rule-Based Hybird Parameters
Abstract
Social media such as Instagram has become an important digital public space for people to voice opinions on various policy issues, including the military education policy, which has recently become highly debated, especially in West Java and even outside Java. The purpose of this research is to develop a sentiment analysis model for public comments on Instagram regarding military education policy using a deep learning approach. m The RoBERTa model was trained and tested using classification performance metrics such as accuracy, precision, recall, and f1-score. The test results show that the model achieved an accuracy of 97%, with the highest f1-score value in the positive category at 0.98. The results show that RoBERTa can effectively classify sentiment based on public opinion on social media. This method can not only provide an overview of public responses, but can also be used as a tool in the decision-making process or public policy evaluation based on real-time digital opinion analysis.
Downloads
Copyright (c) 2025 Jose Julian Hidayat, Cindy Setyowati, Muhammad Dikaisa Ibnu Amin, Khodir Bimasakti, Aditya Pratama Werdana

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.