Development of a Generative Model based Backend of Tutoring Agent

Abstract

In this master thesis, which was accomplished in the Research Seminar of Computer Science in Automotive Software Engineering and whose focus is on designing an advanced backend system for an intelligent tutoring chatbot based on state-of-the-art generative language models. This purpose is realized in the creation of a general generative AI infrastructure to personalize and enhance educational experiences, atscale. This work focuses on the interfacing these models with a chatbot backend to generate natural language responses for questions asked, explanation and personalized feedback. The backend system had to be built out in a way that would allow it to perform important functions like user authentication, session management while users interact with the language model. Datei-Upload durch TIB Moreover, embedding of the tutoring agent in a front-end interface and providing an emotional avatar was reported to have highly improved user engagement and satisfaction. Another example is the emotional avatar that applied NLP to scan emotions of users and then reply with a more sympathetic manner leading to an enhanced support system which was interactive as well. The thesis also includes the creation of a robust dashboard integration to manage and collect user specific datasets. It gives them personalized recommendations and remarked response considering the data it has been collecting, ultimately leading to improving their learning by acknowledging. This thesis is a valuable contribution to educational technology, as it demonstrates one of the many ways generative AI can make personalized adaptive learning experiences possible. The findings from this body of research not only will contribute to continued development of similar intelligent tutoring systems but are also a further steppingstone towards how AI solutions can improve education in practice. Datei-Upload durch TIB