Year: 2025
Type: preprint
Study design: Systematic Review
Abstract: The growth of hate speech on online platforms has highlighted the urgent need for effective detection mechanisms, especially for under-resourced languages like Amharic. This Systematic Literature Review (SLR) examines state-of-the-art techniques used for Amharic hate speech detection, spanning traditional machine learning, deep learning, and hybrid approaches. Insights from ten studies reveal diverse methodologies, including recurrent neural networks (e.g., LSTM, BiLSTM) and multimodal frameworks integrating text and audio features. Despite promising results, challenges persist, such as limited dataset diversity, underexplored feature engineering, and the scalability of advanced models. Evaluation metrics like accuracy and F1-scores vary significantly across studies, with the highest reported accuracy reaching 97.9%. The review underscores the need for larger, more diverse datasets, consistent evaluation frameworks, and deeper error analyses. By identifying research gaps and opportunities, this study aims to guide future advancements in Amharic hate speech detection, ensuring more robust and inclusive solutions for combating online hate speech.
Institutions: Debark University, Woldia University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
SDG: Reduced inequalities
Open Access status: gold
OpenAlex ID: W4413891752