Ανάπτυξη εξατομικευμένου εκπαιδευτικού λογισμικού με χρήση τεχνητής νοημοσύνης και ανάλυσης δεδομένων
Development of personalized software through Artificial Intelligence and data analytics : an educational software perspective
View/
Keywords
Προσαρμοστική μάθηση ; Εξατομικευμένη εκπαίδευση ; Μεγάλα γλωσσικά μοντέλα ; LLM ; Τεχνητή νοημοσύνη ; Εκπαίδευση ; Retrieval-Augmented Generation (RAG) ; Αυτόματη δημιουργία ασκήσεων ; Αυτόματη αξιολόγηση ασκήσεων ; Εξάσκηση κώδικαAbstract
This thesis focuses on the development of an intelligent educational software system that leverages
Artificial Intelligence and data analysis techniques to provide personalized learning in programming.
The main goal is to design a platform capable of adapting to each learner’s knowledge level, skills,
and progress, delivering exercises and feedback tailored to individual needs.
The work begins with a theoretical overview of Personalized Learning Environments and Large
Language Models (LLMs), with particular emphasis on the LLaMA model, which was used in the
implementation. The system’s technical architecture integrates a Flask backend, React frontend,
SQLite database, and FAISS index for semantic retrieval of educational content. Exercise difficulty is
estimated using code complexity analysis through the Radon library, while personalization is based
on per-topic user levels.
The proposed platform supports the generation, evaluation, and adaptive refinement of
programming exercises in real time, aiming to enhance learner autonomy and conceptual
understanding. The results demonstrate that the combination of LLMs with retrieval and data analysis
methods can lead to more efficient and flexible learning environments, contributing to the
advancement of truly personalized education.