Analyzing the Learning Process in a Card Operation-Based Programming Learning Support System

Abstract

The Card Operation-based Programming Learning Support System (COPS) was developed as a tool to assist programming learning with a focus on helping learners understand the structure of programs. Previous studies have demonstrated that when this system is introduced into classes, it maintains learning outcomes equivalent to those of traditional coding-based learning while re-ducing learning time. However, as the difficulty of tasks increases, learners tend to systemati-cally search for answers relying solely on feedback functions (hints) without engaging in thoughtful problem-solving, a phenomenon known as knowledge-free solutions. Therefore, this study aims to enable instructors to intuitively grasp learners engaged in inappropriate learning activities by analyzing and visualizing the learning log data from COPS. Traditionally, the Le-venshtein distance has been used as a quantitative measure to evaluate learners' learning progress. In this study, we aim to enhance the analysis and visualization of the learning process by in-corporating the concept of taboo choices, aiming to improve the accuracy of detecting learners engaged in inappropriate learning activities. Taboo choices are selections with heavier penalties when chosen compared to regular dummy choices. We compared the detection accuracy of inappropriate learners using three patterns of visualization results with different weights for the penalties of taboo choices. The results demonstrate the significance of introducing the taboo choices mechanism in the analysis and visualization process.