A Hybrid Chatbot Model for Enhancing Administrative Support in Education: Comparative Analysis, Integration, and Optimization

The increasing number of students and the limited administrative resources in educational institutions have highlighted inefficiencies in traditional communication methods. Although chatbots offer potential solutions, current implementations are constrained by their architectures: rule-based chatbots lack flexibility, retrieval-based systems depend heavily on predefined datasets, and generative models risk inaccuracies. This study introduces a hybrid chatbot model designed to overcome these challenges by combining rule-based, retrieval-based, and generative approaches. The proposed model employs a Multinomial Naive Bayes classifier to intelligently route user queries to the most appropriate module, ensuring optimized performance across diverse query types. A domain-specific dataset from Sakarya University was developed, featuring 2,253 question-answer pairs categorized into seven administrative domains. Experimental results demonstrate that the hybrid chatbot model outperforms standalone approaches, achieving 97.57% accuracy and improved user satisfaction. Moreover, the model’s flexibility allows seamless adaptation to other sectors, such as healthcare and e-commerce. By significantly reducing administrative workload and enhancing engagement, the proposed hybrid chatbot provides an effective framework for scalable, robust, and adaptable conversational agents in educational settings and beyond.