A Deep Learning Model for Detection of Bias in Amharic News (2024)
Highlights of the study
✍️The research introduces a novel deep learning model using CNNs to detect bias in Amharic news articles, addressing a significant gap in natural language processing resources for less-resourced languages like Amharic.
✍️A diverse dataset of 15,000 Amharic news sentences from social media was meticulously compiled and annotated by domain experts, classifying each article as neutral or biased, which was crucial for training and evaluating the model.
✍️The proposed CNN-based model achieved high accuracy (89.5%) and an F1-score of 80.28% in bias detection on a separate test set, outperforming traditional machine learning techniques and other related works in recall, precision, F1-score, and accuracy for Amharic news bias detection.
✍️This study contributes to improved media literacy and promotes objective reporting by empowering the Amharic-speaking community to critically evaluate news sources, and the developed model can serve as a foundation for bias detection systems in other languages.