Due to significant increases in internet penetration and the development of smartphone technology during the preceding couple of decades, many people have started using social media as a communication platform. Social media has grown to be one of the most significant components, with several benefits. However, technology also poses a number of threats, challenges, and barriers, such as hate speech, disinformation, and fake news. all most all social media platforms can be accused of not doing enough to thwart hate speech on their platform. Amharic and Afaan Oromo language speakers frequently mix the two languages when conversing and posting on social media. 
In this work, a bilingual hate speech detection for Amharic and Afaan Oromo languages were conducted using four different deep learning classifiers (CNN, BiLSTM, CNN-BiLSTM, and BiGRU) and three feature extraction (Keras word embedding, word2vec, and FastText) techniques. According to the experiment result, BiLSTM with FastText feature extraction is an outperforming the other algorithm by achieving a 78.05 % percent of accuracy for Bilingual Amharic Afaan Oromo hate speech detection. Furthermore, we are working towards multilingual hate speech detection for other under-represented Ethiopian languages.

Bilingual Hate Speech Detection on Social Media Amharic and Afaan Oromo

Hate and offensive speech on social media is targeted to attack an individual or group of community based on protected characteristics such as gender, ethnicity, and religion. Hate and offensive speech on social media is a global problem that suffers the community especially, for an under-resourced language like Afaan Oromo language. One of the most widely spoken Cushitic language families is Afaan Oromo. Our objective is to develop and test a model used to detect and classify Afaan Oromo hate speech on social media. We developed numerous models that were used to detect and classify Afaan Oromo hate speech on social media by using different machine learning algorithms (classical, ensemble, and deep learning) with the combination of different feature extraction techniques such as BOW, TF-IDF, word2vec, and Keras Embedding layers. To perform the task, we required Afaan Oromo datasets, but the datasets were unavailable. By concentrating on four thematic areas of hate speech, such as gender, religion, race, and offensive speech, we were able to collect a total of 12,812 posts and comments from Facebook. BiLSTM with pre-trained word2vec feature extraction is an outperformed algorithm that achieves better accuracy of 0.84 and 0.88 for eight classes and two classes, respectively.

Teshome Ababu

2

SHORT BIO