Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

Harnessing Structured Knowledge: A Concept Map-Based Approach for High-Quality Multiple Choice Question Generation with Effective Distractors

Created by
  • Haebom

Author

Nicy Scaria, Silvester John Joseph Kennedy, Diksha Seth, Ananya Thakur, Deepak Subramani

Outline

This paper presents a hierarchical concept map-based framework to address the challenges of generating high-quality multiple-choice questions (MCQs) that target diverse cognitive levels and contain common misconceptions. Targeting high school physics students, we develop a hierarchical concept map that encompasses key physics topics and their interconnections. This concept map is then provided as a structured context for LLMs to generate MCQs and incorrect answers that specifically address misconceptions. An automated validation process ensures the quality of the generated MCQs, and we compare the results with existing LLM-based and RAG-based methods. Expert and student evaluations demonstrate that the proposed method achieves a significantly high success rate (75.20%) and a low guess success rate (28.05%), enabling robust assessment and identification of conceptual gaps across cognitive levels.

Takeaways, Limitations

•
Takeaways:
◦
We present the possibility of automatically generating high-quality MCQs using a hierarchical concept map-based framework.
◦
The possibility of increasing learning effectiveness is presented by creating MCQs that consider various cognitive levels and reflect misconceptions.
◦
Superior performance compared to existing methods verified through expert and student evaluations.
◦
Suggesting the possibility of improving educational efficiency through large-scale assessment and providing individual learning feedback.
•
Limitations:
◦
The current study is limited to high school physics, and further research is needed to determine whether the results can be generalized to other subjects or stages of learning.
◦
Consider the cost and effort involved in developing and maintaining concept maps.
◦
It depends on the performance of LLM, and the limitations of LLM may affect the results.
◦
The evaluation was limited to a specific school and a specific student population, so further research is needed to determine generalizability.
👍
Made with Slashpage