STANCE DETECTION FOR AFAAN OROMO TEXT USING DEEP LEARNING APPROACH

The Afaan Oromo, often known as Oromo, is a significant African language that is spoken by the Oromo people in Ethiopia and some parts of Kenya. It is easy to determine its precise position among African languages since several factors, like the number of native speakers, geographic distribution, and cultural significance, cannot have a distinct impact on rankings. Stance Detection is a task to automatically identify whether a particular news headline “Agrees” with, “Disagrees” with, “Discusses,” or is Unrelated to a particular news article. So, in this study, we proposed to utilize the Deep learning algorithm (LSTM, BiLSTM and GRU) model with different feature extraction on Afaan Oromo text Stance Detection. Because, feature extraction is useful to reduce dimensions and get the best feature from the dataset by selecting and combining variables into features, thus, effectively reducing the amount of data. In this research, we have used 12528 a newly collected and annotated pair of headline and body of Afaan Oromo text dataset to develop pre-trained word embedding model. The data train test splits (80/20) were used to choose the model that performed the best out of all the algorithms. Also, the text preprocessing activities of normalization, tokenization, text cleaning, and stop word removal were just a few of the NLP tasks that were carried out in this work. In our experimentation, in terms of accuracy, precision, recall, and F1-score, the models created by the LSTM, BILSTM, and GRU algorithms produced more promising outcomes than other models. The results show that ordinary LSTM and GRU based modeling performs less than BILSTM-based modeling. BILSTM has great accuracy with 86.63% accuracy, 87% Precision, 89% recall, 87% f1-scores for the BILSTM model respectively, indicating the higher model's performance. Key-words: Stance detection, Afan Oromo Language, Deep learning, Social Media