Analisis Efektivitas Algoritma Machine Learning dalam Deteksi Hoaks: Pada Berita Digital Berbahasa Indonesia

Abstract

The rapid development of information technology has transformed how society accesses and disseminates information. Unfortunately, this phenomenon also creates opportunities for the massive spread of fake news or hoaxes through digital platforms. This research aims to analyze the effectiveness of several machine learning algorithms in detecting text-based hoaxes in Indonesian. The algorithms tested include Multilayer Perceptron (MLP), Naïve Bayes (NB), Support Vector Machine (SVM), and Random Forest (RF). The data used consists of online news articles that have undergone text preprocessing stages such as tokenizing, case folding, filtering, stopword removal, stemming, and weighting using the TF-IDF method with a combination of unigram and bigram features. Performance evaluation was conducted using precision, recall, F1-score, and accuracy metrics. The results show that the SVM and MLP algorithms yielded the highest performance with evaluation values above 99.8%, while RF demonstrated strong and stable performance, and NB showed decent performance with high efficiency. These findings provide insights into the effectiveness of text classification methods in hoax detection and serve as a reference for developing more efficient and accurate fake news detection systems