Building and Evaluating a Chatbot Using a University FAQs Dataset

The integration of artificial intelligence in educational settings has increasingly shown potential to enhance the efficiency and accessibility of information. One such application is the development of chatbots that can provide instant responses to frequently asked questions, thereby alleviating the workload on administrative staff and improving user experience. This paper presents the development and evaluation of a chatbot designed to assist students and staff at a university by providing accurate responses to common queries. Using a dataset sourced from Kaggle, the chatbot is trained on various intents, including course information, fees, hostel facilities, and more. The dataset undergoes thorough preprocessing, including tokenization, lemmatization, and vectorization to ensure effective model training. The neural network model is built using TensorFlow and comprises multiple dense layers with ReLU activation functions and dropout layers to prevent overfitting. The model is trained over 200 epochs with a batch size of 5, utilizing the Adam optimizer and categorical cross-entropy loss function. The results demonstrate the chatbot’s high accuracy and effectiveness, achieving an accuracy of 99.75%, precision of 99.76%, recall of 99.75%, and an F1 score of 99.75%. These metrics indicate the model’s robustness in understanding and responding to user queries accurately. The implications of this study suggest that implementing such chatbots in educational institutions can significantly streamline information dissemination and improve user engagement.