Perbandingan Algoritma Support Vector Machine dan Bi-Directional Long Short Term Memory Dalam Mengklasifikasi Berita Hoaks

Abstract

The rapid advancement of digital technology has made it easier to spread information widely and quickly. However, this ease of access has also contributed to the rise of false or misleading news, commonly known as hoaxes, which can confuse the public. This study aims to evaluate and compare the performance of two machine learning algorithms, namely Support Vector Machine (SVM) and Bi-Directional Long Short Term Memory (BiLSTM), in classifying hoax news written in Indonesian. The research adopts a supervised learning approach, where models are trained using pre-labeled data categorized as either hoax or non-hoax. The process begins with collecting data from trusted sources, followed by several preprocessing steps, including text cleaning, tokenization, stopword removal, and stemming. After preprocessing, the dataset is split into training and testing sets in an 80:20 ratio. The results show that the SVM model achieved an accuracy of 98.46%, with 98% precision and 99% recall for the non-hoax category. In comparison, the BiLSTM model performed better, reaching 99% accuracy, with both precision and recall at 99% for both categories. These findings indicate that BiLSTM is more effective at capturing linguistic context and identifying patterns in hoax-related content. Additionally, the models were implemented into a web-based system to assess their real-world detection capabilities.