AGRIRESPONSE: A REAL-TIME INTELLIGENT QUERY SYSTEM FOR NATIONWIDE AGRICULTURAL ASSISTANCE

Abstract

Information science developments may be very helpful in advancing the country's sustainable agricultural objectives. In order to address the need for prompt assistance for Indian farmers throughout the country, we provide a framework for a text-based query-response generation system. Building a knowledge base that can respond to enquiries about plant protection from a wide range of farmers is one of the main problems in developing such systems. The call logs from the nationwide farmers' helpline network over the previous eight years are gathered and analysed to provide the necessary information foundation in order to address this issue. Furthermore, in order to administrate the user-input queries and extract pertinent responses from the database, three response-retrieval models with approximate matching and spatial-based searching capability are constructed. A varied question bank with 755 questions spanning 151 Indian crops is created in order to verify the effectiveness of the suggested structure. The models are evaluated using three metrics: Average Response-retrieval time, Crop-weighted Performance Score, and Accuracy Percentage. According to experimental findings, AgriResponse is a helpful framework for real-world applications, with various retrieval models being appropriate for various situations.