Twitter Sentiment Analysis to Assess Public Opinion on Jokowi's Performance Over Two Periods using the Recurrent Neural Network (RNN) Method

Abstract

This study aims to analyze public sentiment toward President Joko Widodo’s performance over his two terms in office by utilizing data from the social media platform Twitter. Twitter was chosen due to its real-time nature and popularity among users for expressing opinions openly. The method employed is a Recurrent Neural Network (RNN) with a Long Short-Term Memory (LSTM) architecture, which is well-suited for processing sequential text data. The research process involves several preprocessing stages, including text cleaning, normalization, tokenization, stopword removal, and stemming, to prepare the textual data for model input. The dataset consists of 1,006 Indonesian-language tweets categorized into three sentiment classes: positive, negative, and neutral. After simplifying the classification into two classes, the model achieved an Accuracy of 63.33%. Evaluation using a Confusion matrix and classification report indicates that the model performs better in identifying positive sentiment, though it still struggles with misclassifying negative sentiment. The results demonstrate that the RNN method is fairly effective for social media-based sentiment analysis. It is expected that this study can serve as a reference for government institutions or policy researchers to understand public perceptions and support more data-driven and responsive policymaking processes.