자연어 기반 콘텐츠 추천 시스템의 효과성 분석에 관한 연구: 초등학생 대상 소프트웨어 교육의 사례를 중심으로

Study on the Effectiveness Analysis of Natural Language-Based Content Recommendation System in Elementary SW Education

This study aims to evaluate the educational effectiveness of a natural language-based content recommendation system designed for elementary software (SW) education. The proposed system recommends personalized learning content by analyzing students’ free-text queries and Entry Test results. It utilizes a dual recommendation mechanism: one based on semantic similarity using a pre-trained KoBERT model, and the other based on meta-learning through a Model-Agnostic Meta-Learning (MAML) algorithm. To assess its impact, a total of 40 elementary school students participated in a 4-session program, during which they interacted with the recommendation chatbot and completed assigned coding tasks. Both pre/post-intervention surveys were conducted to measure changes in learning motivation, engagement, and perceived self-efficacy. The results indicate that students exhibited increased interest and self-directed learning behavior, particularly when recommendations were aligned with their input queries and prior knowledge. The findings support the effectiveness of a natural language-based personalized recommendation approach in fostering engagement and adaptive learning in elementary SW education.