Iwe is a research project looking to address the depth of artificial intelligence in informal education using a native language as a variable. The bulk of LLMs are trained in global languages and equipped with the knowledge to perform basic tasks, this technology has revolutionised mankind's approach to mundane matters and has earned a reputation as a guide to more complex tasks. In our context we want to apply the technology to Yoruba and include a Router System to streamline knowledge flow and improve expertise in our models, making it more accessible and fostering the preservation of culture. We want to enhance how we engage with literature, providing a unique and supportive learning experience.
At Iwe we are building a comprehensive digital library of yoruba literature encomprising of classical and contemporary yoruba fiction and non-fiction. A collection of poetry from historical to modern times. Scholarly papers, essays of the yoruba socio-political landscape and digitized folktales and proverbs passed down through generations. Chapter based reading summarizing full texts into interactive segments and adaptative reading.
Discover a vast collection of authentic Yoruba literature, including books, journals, and articles sourced from multiple repositories. Explore original texts or access translated versions for a broader understanding of Yoruba culture.
To support accessibility and inclusivity, Iwe incorporates Text-to-Speech functionality to enable users listen to books and articles read in a natural Yoruba voice with a dialect of choice, together with our Speech-to-Text functionality enabling users with diction-based interactions.
Iwe integrates a RAG (Retrieval Augumented Generation) system to enable direct interaction with texts, users can ask questions about books, e-journals and receive responses sourced from original texts.
HOW IT WORKS
The data centre forms the engine of the system. Here, data is extracted from various preselected sources, cleaned and transformed into elements that are suitable for Machine Learning and Natural Language Processing, this is where we store and index Yoruba content and crunch a suitable output for user enquires, our models are trained on specific data to ease load shedding and be cost efficient. This component serves as the brain of the system.
The RAG (Retrieval Augumented Generative) system forms a bond between the user and content, showcasing generative AI capabilities, users have the liberty to interact with yoruba texts, ask questions about books and receive responses sourced from the knowledge base. Micro-learning assistance is provided in circumstances where users can explore meanings, historical contexts and insights about specific words. phrases, passages and traditions. Feeding off the knowledge base of the data centre, exercises and quizzes can be generated from chosen contents.
Our Text-to-Speech (TTS) and Speech-to-Text (STT) supports accessibility and inclusivity. This service encourages diction-based interactions together with translation capabilities. research and development efforts focus on incorporating emerging technologies like voice recognition.