Bridging the gap in college information access through natural language processing powered Lucy chatbot

Abstract

In recent years, there has been a notable shift in how people interact with technology, marked by the growing popularity of voice assistants and chatbots. As these systems’ capabilities improve, they are increasingly becoming the preferred method of contact between customers and technology. Chatbots are popular owing to their simplicity and efficiency. Chatbots may manage several users’ inquiries simultaneously, removing the need for individual staff members to address them. Therefore, businesses that integrate chatbots into their operational systems save money. This study empirically investigated the efficacy of natural language processing (NLP) in creating a chatbot specifically tailored to handle college information inquiries. This paper introduces Lucy, a chatbot that utilizes a deep neural network in conjunction with complex language models including BERT, RoBERTa, and DistilBERT architectures. College students were surveyed to assess Lucy’s performance in a user research study. The research showed that Lucy is effective as a teaching tool for imparting knowledge to students, with an accuracy rate above 85%. From the evidence, Lucy is capable of subsequently reducing in-person enquiries by acting as a centralized forum for student questions, which highlights the chatbots in the education sector and also demonstrates the capabilities to improve the quality of information shared to the students.