Claim Missing Document
Check
Articles

Found 1 Documents
Search
Journal : Journal Of Artificial Intelligence And Software Engineering

Enhancing Hoax News Detection Performance through Modified Convolutional Neural Network Architecture Mujilahwati, Siti; Reknadi, Danang Bagus
Journal of Artificial Intelligence and Software Engineering Vol 5, No 2 (2025): Juni On-Progress
Publisher : Politeknik Negeri Lhokseumawe

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30811/jaise.v5i2.6953

Abstract

The spread of fake news (hoaxes) on the internet is increasingly widespread, along with the increasing use of the internet in Indonesia. To detect hoaxes automatically, this study proposes a Convolutional Neural Network (CNN) model based on data from the Detik.com and TurnBackHoax.id sites. The data includes various categories such as politics, religion, and technology. The model was developed through several optimization stages: tokenization and text padding, utilization of pre-trained Word2Vec as embedding weights, stratified data division, handling class imbalance with class weights, and implementing a multilevel CNN architecture with dropout. Training was carried out using the Adam optimizer with a small learning rate and an early stopping technique. The evaluation results showed that the proposed model achieved an accuracy, precision, recall, and F1-score of 96%, with the fastest training time of 100.42 seconds. The model was also evaluated with the ROC_AUC model, with a score of 98.91%. This performance outperformed the CNN-1D and Augmentasi-CNN models. This approach has proven to be effective and efficient in detecting Indonesian hoax news.