Extractive Question Answering Over Ancient Scriptures Texts Using Generative AI and Natural Language Processing Techniques

Generative AI (GenAI) and Natural Language Processing (NLP) are transforming the landscape of question-answering systems, shifting from traditional methods to more advanced, automated solutions. This study explores into applying these technologies using a unique dataset derived from the ancient Hindu scriptures, specifically the Atharv Ved. Employing a combination of vector embedding transformations and a novel question-answering model integrating BERT embeddings, attention mechanisms, and Gated Recurrent Unit (GRU) layers, we have developed a system capable of understanding and generating semantically rich questions and answers. Our model was rigorously evaluated against the Atharv Ved and SQuAD datasets, demonstrating high accuracy rates of 94% and 95%, respectively. These results underscore the model’s capability to handle complex, context-heavy textual data effectively. The study concludes with the model’s potential applications in digital humanities, preserving cultural heritage, and enhancing the accessibility of ancient texts through modern technology.