Development of AI-BasReinforcement Learnined Academic Chatbot for Higher Education Student Services

Abstract

This study focuses on the development of an AI-based academic chatbot designed to enhance the efficiency and responsiveness of university administrative services. The increasing demand for fast, accurate, and accessible academic information has created challenges for conventional service systems that rely heavily on human staff and manual operations. To address this issue, the research aims to design and evaluate a chatbot system that integrates Natural Language Processing (NLP) for understanding user input and Reinforcement Learning (RL) for continuous performance improvement based on feedback. The research employs a Design and Development Research (DDR) approach with system modules comprising text input, NLP engine, decision module with RL, and response output. The chatbot was trained using a dataset of academic queries and evaluated through three main metrics: accuracy, response time, and user satisfaction. The findings indicate that the chatbot achieved an accuracy rate of 85.2%, exceeding the target of 85%, with an average response time of 2.4 seconds—faster than the 3-second benchmark. Additionally, user satisfaction reached 84.3%, demonstrating the system’s effectiveness and usability. The overall results reveal that the combination of NLP and RL provides significant improvements in chatbot responsiveness and adaptability, enabling more human-like and contextually appropriate interactions. This study concludes that AI-driven chatbots can serve as an effective and scalable digital solution for academic service automation, reducing administrative workload and improving student engagement. Future research is recommended to enhance linguistic diversity, integrate multimodal communication, and implement continuous reinforcement learning to achieve more natural and intelligent academic interactions.