Natural Language Processing Frontiers in Djiboutian African Languages

Natural Language Processing (NLP) is a field within Computer Science that aims to enable machines to understand and process human language effectively. Despite its widespread application in English and other major languages, NLP for African languages has been underexplored. The methodology involves an extensive literature review to identify existing NLP research for African languages, followed by an empirical analysis using state-of-the-art machine learning models tailored for NLP tasks such as sentiment analysis and named entity recognition. A key aspect of the work is the development and evaluation of custom language-specific models. A significant finding was the variability in performance across different Djiboutian African languages, with some models achieving up to 85% accuracy in sentiment classification for Tigrinya, while others struggled with lower accuracies around 60%. This highlights the need for tailored approaches and increased data availability. The study concludes that while there are substantial challenges in developing NLP systems for African languages, these can be overcome through targeted research and collaboration. The development of custom models is crucial to achieving high accuracy across diverse linguistic contexts. Future work should prioritise the collection of more annotated data for underrepresented languages and explore hybrid approaches combining traditional machine learning with contemporary deep learning techniques. Natural Language Processing, African Languages, Djibouti, Custom Models, Sentiment Analysis Model estimation used $\hat{\theta}=argmin_{\theta}\sum_i\ell(y_i,f_\theta(x_i))+\lambda\lVert\theta\rVert_2^2$, with performance evaluated using out-of-sample error.