NLP Chatbot with ANN and SGD for University Admissions Website

Abstract

Universities today face challenges in providing real-time and responsive information for New Student Admissions. This research aims to implement a chatbot based on Natural Language Processing (NLP) using an Artificial Neural Network (ANN) model optimized with Stochastic Gradient Descent (SGD) to address these challenges. The goal of the implementation is to enhance information services through the use of chatbots. The analysis was conducted on frequently asked questions (FAQs) from prospective students, followed by system design, ANN architecture development, and model performance testing. During the ANN model development, Term Frequency–Inverse Document Frequency (TF-IDF) was employed for numerical representation, enabling the training process to be processed effectively by the ANN model. The implementation results show that the chatbot deployed on the UBHINUS admission website successfully provides real-time and relevant information. Testing results demonstrate an accuracy of 92.5% and an F1-Score of 91.56%, indicating strong classification performance. It can be concluded that the chatbot is effective in facilitating access to information regarding New Student Admissions at UBHINUS.