Design and Implementation of an AI-Driven Quiz Generator
Al-gburi, Mohammed (2025)
2025
Master's Programme in Computing Sciences and Electrical Engineering
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2025-06-10
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202506107054
https://urn.fi/URN:NBN:fi:tuni-202506107054
Tiivistelmä
The increasing capabilities of Large Language Models such as GPT-4 and Meta’s Llama have enabled new opportunities in personalized and automated learning experiences. This thesis introduces APUOPE-RE, a web-based platform designed to assist students in the requirements engineering course by generating quizzes from provided materials using LLMs. APUOPE-RE aims to support self-directed learning and active recall by automating the creation of context-relevant quizzes.
The system supports two question formats: single-choice questions (SCQs) and true/false questions. It dynamically adjusts the difficulty of generated questions in response to user performance, delivering a personalized and adaptive learning experience. Additionally, it enables users to review and track their past quizzes, supporting ongoing self-assessment and progress monitoring.
This thesis investigates the effectiveness of large language models in enhancing quiz generation and explores the technical challenges involved in implementing such a system. Effectiveness was addressed through user feedback from eight participants. Their responses indicated that the AI-generated quizzes were generally accurate, relevant, and easy to use, with most users reporting satisfaction in terms of content quality and interface design. While some challenges arise in implementing LLM-powered quizzes, key issues included model integration, prompt engineering, latency, and maintaining output consistency. Overall, these findings suggest that while LLMs hold significant promises for educational applications, successful deployment demands careful design, ongoing refinement, and technical resilience.
The system supports two question formats: single-choice questions (SCQs) and true/false questions. It dynamically adjusts the difficulty of generated questions in response to user performance, delivering a personalized and adaptive learning experience. Additionally, it enables users to review and track their past quizzes, supporting ongoing self-assessment and progress monitoring.
This thesis investigates the effectiveness of large language models in enhancing quiz generation and explores the technical challenges involved in implementing such a system. Effectiveness was addressed through user feedback from eight participants. Their responses indicated that the AI-generated quizzes were generally accurate, relevant, and easy to use, with most users reporting satisfaction in terms of content quality and interface design. While some challenges arise in implementing LLM-powered quizzes, key issues included model integration, prompt engineering, latency, and maintaining output consistency. Overall, these findings suggest that while LLMs hold significant promises for educational applications, successful deployment demands careful design, ongoing refinement, and technical resilience.