Comparative Analysis of Deep Learning Based Afaan Oromo Hate Speech Detection

Abstract The network and openness of online media stages permit individuals to communicate their thoughts and offer encounters without any problem. Nonetheless, the Internet, as a typical stage for correspondence, has been weaponized to spread a blend of misrepresentations, misuse, and inclinations with far-running outcomes. Prominently, disdain discourse influences society in numerous angles, for example, influencing the emotional well-being of designated crowds, influencing social association , and prompting viciousness and interruption of properties. Whenever left unchecked, disdain discourse can possibly sabotage confidence in electing processes and undermine popularity-based foundations. Web-based media goliaths, for example, Facebook have been scrutinized for permitting its foundation to be utilized as a vehicle for disdain discourse , with least thought given to non-English pages, which has led to intermittent issues in strongly enraptured nations like Ethiopia and other emerging nations.Subsequently, there is a need to construct a hate speech detection model for Ethiopian languages like Afaan Oromo. Aiming to calm the effect of hate speech this paper examines the viability of deep learning models for Afaan Oromo hate speech recognition. Toward this, we collect and annotate the first and largest Afaan Oromo social media datasets. Variations of profound deep learning models such CNN, LSTMs, BiLSTMs, LSTM, GRU, and CNN-LSTM are examined to evaluate their viability in identifying Afaan Oromo Hate speeches. The examination result uncovers that the model dependent on CNN and Bi-LSTM somewhat outperforms every one of the models on the test dataset with an average F1-score of 87%. Overall, considering the Article Title nature of the Afaan Oromo language and the prevalence of hate speech, we believe that the finding of this study is promising for future works.