Agricultural Question Answering Dataset Creation and its Evaluation through BERT based Experimental Analysis

In the field of Information Retrieval (IR). Question Answering (QA) is considered as a contemporary area for providing relevant answers in response to questions that are proposed in natural language. The QA system consist of three core components and these are question classification, information retrieval and answer extraction. The goal of QA is to develop automated systems that can comprehend the meaning of questions and provide either accurate or relevant answers. Machine learning algorithms plays a vital role in solving automated question answering (QA) problems using natural language processing. These algorithms are designed to enable computers to understand, infer, and respond to human language queries. This work considered deep learning based LLM-BERT for implementing a QA system over the agricultural domain. BERT is used because it comprehends the context of words by observing at both their left and right environment, making it highly operative for complex question and answers. Its ability to adapt to specific domains through fine-tuning ensures accurate and relevant responses in tasks like QA. The dataset used here has been collected, generated and constructed by the authors.