Abstract: Hate speech on social media platforms like YouTube, Facebook, and Twitter threatens online safety and societal harmony. Addressing this global challenge requires innovative and efficient solutions. We propose DBFN-J (DistillBERT-Feedforward Neural Network with Jaya optimization), a lightweight and effective algorithm for detecting hate speech. This method combines DistillBERT, a distilled version of the Bidirectional Encoder Representations from Transformers (BERT), with a Feedforward Neural Network. The Jaya algorithm is employed for parameter optimization, while aspect-based sentiment analysis further enhances model performance and computational efficiency. DBFN-J demonstrates significant improvements over existing methods such as CNN BERT (Convolutional Neural Network BERT), BERT-LSTM (Long Short-Term Memory), and ELMo (Embeddings from Language Models). Extensive experiments reveal exceptional results, including an AUC (Area Under the Curve) of 0.99, a log loss of 0.06, and a balanced F1-score of 0.95. These metrics underscore its robust ability to identify abusive content effectively and efficiently. Statistical analysis further confirms its precision (0.98) and recall, making it a reliable tool for detecting hate speech across diverse social media platforms. By outperforming traditional algorithms in both performance and resource utilization, DBFN-J establishes a new benchmark for hate speech detection. Its lightweight design ensures suitability for large-scale, resource-constrained applications. This research provides a robust framework for protecting online environments, fostering healthier digital spaces, and mitigating the societal harm caused by hate speech.

Nourah Fahad Janbi, Abdulwahab Ali Almazroi and Nasir Ayub, “DBFN-J: A Lightweight and Efficient Model for Hate Speech Detection on Social Media Platforms” International Journal of Advanced Computer Science and Applications(IJACSA), 16(1), 2025. http://dx.doi.org/10.14569/IJACSA.2025.01601128

@article{Janbi2025,
title = {DBFN-J: A Lightweight and Efficient Model for Hate Speech Detection on Social Media Platforms},
journal = {International Journal of Advanced Computer Science and Applications},
doi = {10.14569/IJACSA.2025.01601128},
url = {http://dx.doi.org/10.14569/IJACSA.2025.01601128},
year = {2025},
publisher = {The Science and Information Organization},
volume = {16},
number = {1},
author = {Nourah Fahad Janbi and Abdulwahab Ali Almazroi and Nasir Ayub}
}