Use of LLM for personalized module schedule planning
The planning of studies and semesters is of great importance for students. This process is, however, a time-consuming task. For conscientious planning, several module descriptions must be compared individually and checked for relevance to the student's individual needs and goals. In the past, various methods such as recommendation systems have been used to support students in their individual planning. However, such tools offer limited opportunities to consider personal preferences.
Arunasalam, Anojan, 2025
Art der Arbeit Master Thesis
Auftraggebende
Betreuende Dozierende Hinkelmann, Knut, Spahic, Maja
Views: 2
Large Language Models (LLM) offer a promising alternative, as they are easy to use and allow students to interact directly with the system and therefore shape the planning according to their preferences. At the same time the planning capability of LLMs has been critically examined in the literature. It has been shown that LLMs alone are not able to generate functioning plans, but they can be used as an instrument that supports planning if it is expanded with other instruments.
This paper therefore attempts to combine LLMs with a Retrieval Augmented Generation (RAG) component to develop a artifact that generates working study and semester plans.
The qualitative evaluation of the developed artifact has shown that functioning semester plans can be created successfully in a human-in-the-loop approach. Holistic study plans that take into account different restrictions and requirements can also be created. However, there is currently a lack of medium to long-term information, such as the final timetables, which is necessary to ensure a truly functioning plan.
Studiengang: Business Information Systems (Master)
Keywords
Vertraulichkeit: öffentlich