Publication

Abstract : Agriculture plays a vital role in the economy of many nations, especially in regions where a significant portion of the population depends on it for their livelihood. Despite its importance, linguistic and technological barriers often hinder access to accurate agricultural information. This article introduces a multilingual chatbot designed to address agricultural queries efficiently. Leveraging Natural Language Processing and deep learning models, this multi-language chatbot accommodates English, Hindi, Malayalam, Kannada, and Telugu. Since the query is matched with relevant responses using cosine similarity on the sentence-transformers architecture, the system ensures that relevant responses are given. With IndicTrans2 (AI4Bharat), a robust neural machine translation model tailored for Indian languages, the facility provides a single-machine foundation on which translations about 22 Indian languages with subsequent additional features-text-to-speech functionality enhances accessibility to varied users. The system design, data preprocessing, implementation, and evaluation metrics are covered in this study, which also shows how it might help farmers and other agricultural stakeholders.