Leveraging Technology to Empower Millet Farmers: A Retrieval-Augmented Generation Approach with Large Language Models

This research addresses the critical need to bridge the gap between traditional agricultural knowledge and modern findings, particularly focusing on millet cultivation in India. Millet, being a staple grain in the region, holds significant importance for both farmers and agricultural researchers. However, the lack of accessible technology exacerbates the dissemination of crucial information to farmers. To tackle this challenge, our study proposes a comprehensive solution leveraging cutting-edge technologies such as Large Language Models (LLMs), Prompting, and vector database management. The core of our approach centers on the utilization of RAG techniques, which involves scraping and segmenting millet-related data into manageable chunks. These chunks are then transformed into vectors using LLMs and stored in the vector database. Similarity measures such as cosine similarity to obtain relevant vectors in response to questions posed by farmers. Prompting is then employed to provide responses that understood by humans. Crucially, the technology prevents the spread of false information by refusing to respond to questions that are not related to agriculture, thus guaranteeing accuracy.