Designing A Llama 2-Powered Chatbot for Enhanced College Website Support
R. Thamilselvan, P. Natesan, A Manimaran, S E Naveenkumar, J K Shanthosh, S. Vigneshwaran · 2024
Most of the Colleges and universities are increasingly using artificial intelligence (AI) to improve website support and user experience in the fast-paced digital landscape of higher education. This research investigates a Natural Language Processing (NLP) based chatbot for college website support. In higher education’s digital landscape, AI-driven chatbots, particularly using the Large Language Model Meta AI Version 2 (LLaMA 2) model, are enhancing website support and user experience by addressing inquiries on admissions, courses, resources, and campus life. Another subset of research component is mainly focusing on predicting the cut-off marks for students for joining the institutions. Focusing on student/institutional cut-off mark prediction, the chatbot employs four different regression models to consider factors like joining year, quotas, previous marks, department preferences, exam difficulty, and job market demand. Evaluation metrics such as MSE, RMSE, and R-squared gauge predictive accuracy. By analyzing historical data and emphasizing academic performance, the study determines which regression models best predict student cut-off marks, ensuring a data-driven approach to decision-making. This AI-powered chatbot enhances user experience by providing comprehensive answers and accurately forecasting students’ academic performance.