Development and Evaluation of a Bidirectional LSTM Classifier for Sentiment Analysis in the Hausa Language

Sentiment analysis, which is crucial for gaining insight into public opinion, presents substantial difficulties when used with under-represented languages like Hausa, which is spoken by more than 50 million people globally. The generic machine learning models used by traditional methods sometimes struggle to handle the linguistic intricacies and code-mixing inherent in Hausa text. In this study, we specifically create and assess a bidirectional long short-term memory (Bi-LSTM) classifier for Hausa sentiment analysis. We use sophisticated preprocessing methods and word embeddings for feature extraction by utilising the NaijaSenti dataset, which comprises over 30,000 human-annotated Hausa tweets with a significant amount of code-mixed content. In terms of sentiment categorisation, the Bi-LSTM model fared better than traditional models like Gaussian naive bayes, logistic regression, and random forest. This research addresses a critical gap in natural language processing for Hausa, with broad applications in social media analysis, market trends, and sociolinguistics, ultimately contributing to more inclusive and diverse language technology.