Penggunaan Deep Learning untuk Mengklasifikasi Hate speech dan Good Speech Terhadap Pertamina di Platform Twitter dengan Metode Convolutional Neural Network (CNN)

Abstrak

Advances in digital technology have changed the way people express their opinions, particularly through social media platforms like Twitter. In social and corporate contexts, Pertamina, a state-owned energy company in Indonesia, is frequently the subject of public discourse, both in the form of positive (good speech) and negative (hate speech) expressions. To manage this information, a system capable of accurately and automatically classifying tweets is crucial. This study aims to develop a Deep Learning-based text classification model, specifically using the Convolutional Neural Network (CNN) method, to identify tweets containing hate speech and good speech related to Pertamina. Data was collected from Twitter using relevant keywords, followed by manual preprocessing and labeling. The cleaned dataset was then divided into training and testing data for processing using a CNN architecture. The results showed that the CNN model performed very well in the classification task, achieving a validation accuracy of 98.08% and a testing accuracy of 97.66%. Evaluation using a confusion matrix also showed high precision and recall values, with an f1 score of 0.98. These findings demonstrate that the CNN method is effective in accurately identifying hate speech and positive speech in Indonesian text data, particularly regarding issues related to Pertamina. This research is expected to contribute to the development of automated social media monitoring systems and public opinion management tools.