thedavidemmanuel/Hausa-English-Translation-Model

## English to Hausa Translation Model with RNNs and Attention This repository contains a project that demonstrates the development of an English to Hausa translation model using Recurrent Neural Networks (RNNs) with an attention mechanism. The model is built using TensorFlow and Keras and is trained on a dataset of parallel English-Hausa sentences collected from Twitter. ## Table of Contents - Introduction - Dataset - Data Preparation - Exploratory Data Analysis (EDA) - Model Architecture - Encoder - Decoder with Attention - Training - Evaluation - Attention Mechanism Visualization - Word Embedding Visualization - Challenges and Solutions - Conclusion and Future Work - Dependencies - Usage - Acknowledgments ## Introduction Machine translation is a crucial application of Natural Language Processing (NLP) that automates the translation of text or speech from one language to another. This project focuses on developing a neural machine translation model capable of translating sentences from English to Hausa, a Chadic language widely spoken in West Africa. Despite the limited resources and tools available for Hausa-English translation, this project leverages the power of RNNs and attention mechanisms to handle the complexities of language translation, particularly for less-resourced languages like Hausa. ## Dataset ### Source The dataset used in this project is a collection of parallel English-Hausa sentences extracted from Twitter. It includes both the original tweets and their replies, providing a diverse set of sentence structures and contexts. - **GitHub Repository**: Hausa Corpus by Isa Inuwa-Dutse. - **Dataset File**: `parallel-hausa-tweets.csv`. ### Columns Used - **CleanedMainT**: Hausa main text. - **Hausa2EngMainT**: English translation of the main text. - **CleanedReplyT**: Hausa reply text. - **Hausa2EngReplyT**: English translation of the reply text. _Note: The dataset is included in this repository under the `data/` directory for ease of replic …