A Crop Recommendation System Using Transformer-Based NLP Models

Traditional crop recommendation systems are of-ten limited in analyzing diverse environmental parameters and adapting to new data patterns. In this work, RoBERTa, ALBERT and DistilBERT are utilized to develop a crop recommendation system. A benchmark data set, which contains key agricultural characteristics such as soil nutrients (N, P, K), temperature, humidity, pH, and rainfall, is used to model crop suitability. These models are used to capture the interactions between environmen-tal variables and crop viability. An interactive chatbot interface is developed to provide real-time crop recommendations based on user-inputted conditions. The system is designed to improve accessibility and usability, allowing agriculture stakeholders to make informed decisions.