Akademik Veri Yönetim Sistemi
ICITS 2025, Ankara, Türkiye, 18 - 20 Eylül 2025, ss.122, (Özet Bildiri)
This study presents the development and implementation of an AI chatbot utilizing a fine-tuned large language model (LLM) for enhancing educational services in a Moodle Learning Management System (LMS). An open-source platform, Anything LLM, was deployed on a dedicated server to create a customizable training environment. Data pertaining to the remote education services provided by İnönü University's Distance Education Unit was collected and used to fine-tune the LLM employing Retrieval-Augmented Generation (RAG) techniques, enabling the model to deliver contextually relevant and accurate responses tailored to university-specific queries. Subsequently, a Moodle Chatbot Plugin was developed using scripts generated by the trained workspace and integrated into İnönü University's Moodle LMS portals. The effectiveness of the chatbot was evaluated through analysis of student chat history data and satisfaction surveys, which indicated high efficiency and user satisfaction. Additionally, a notable reduction in support requests to the education unit was observed, demonstrating the chatbot's practical impact on operational efficiency. This approach highlights the potential of fine-tuned LLMs in educational AI applications, offering a scalable solution for personalized student support in e-learning environments.