PERSONALIZED MCQ GENERATOR FOR ADAPTIVE LEARNING BY USING LEVERAGING RAG METHOD IN GEN AI

Abstract

In the development of more automated and adaptive learning environments, a large part of it is to use artificial intelligence in machine-generated multiple-choice questions (MCQs) generated from a combination of different sources. Therefore, we propose a novel intelligent Quiz Generator with which an educator can automatically generate multiple choice questions (MCQs) with high quality questions as well as contextually relevant and challenging questions. In the proposed system, we propose large language model with high quality questions and retrieval augmented generation (RAG) and some optimization techniques for prompt engineering for generating very high quality and contextually relevant quiz questions. A chain-of-thought and self-refine prompting techniques are applied for enabling the model to generate questions that not only are consistent with learning objective but also take into account the common student misconceptions. In addition, we apply performance matrix to support the analysis of individual learning progress of each student. The matrix takes into consideration the results of previous questions, which finds areas of strength and weaknesses, and adjusts future quizzes accordingly to foster accurate discovery of learning needs among students. Based on extensive evaluation we could conclude that the automated system can generate various and adaptive MCQs with higher engagement and comprehension of students. With the aid of artificial intelligence and adding performance analytics, this project provides a reliable and innovative solution for automatic quiz generation and moving forward Personalized.