Implementing Bidirectional Encoder Representations fromTransformers and Gated Recurrent Units in a Chatbot for Student Support Services

Establishing efficient communication among students, faculty, and administrators is imperative ‎to the effective operation of educational institutions. Understanding students’ expectations for ‎fast, flexible, and personalized access to support services can improve satisfaction and strengthen ‎institutional performance. The increasing demand for prompt solutions in question–and–answer ‎services has led to the emergence of intelligent automation, particularly AI–powered chatbots, to ‎address existing educational challenges. This study developed and evaluated a web–based ‎student support system featuring a chatbot that integrates Bidirectional Encoder Representations ‎from Transformers (BERT) and Gated Recurrent Units (GRU) to enhance student support ‎services. The NLP–driven chatbot provides real–time, automated responses to student inquiries, ‎such as enrollment, schedules, tuition, and academic policies. Experimental results showed that ‎the BERT model effectively identified user intent, achieving precision, recall, and F1 scores of ‎‎0.87 and above. The GRU–based decoder generated coherent and contextually relevant replies, ‎validated through ROUGE metrics. Usability testing based on ISO 9241–11:2018 yielded mean ‎scores above 4.5, confirming the system’s effectiveness and user satisfaction. Overall, the ‎BERT–GRU chatbot demonstrates a practical and scalable solution for providing 24/7 student ‎assistance and improving institutional support. Recommendations were proposed to enhance the ‎implementation, effectiveness, and future development of BERT–GRU chatbot systems in ‎educational settings‎.