Khung AI-MSC lập hồ sơ sáng tạo của sinh viên ngành Giáo dục mầm non: Đề xuất và kiểm thử trên dữ liệu mô phỏng

Tóm tắt

In the current context of digital transformation, assessing student creativity requires leveraging multiple sources of learning evidence rather than relying solely on grades or point-in-time assessments. This study proposes the AI-MSC (Artificial Intelligence for Measuring Student Creativity) framework to profile the creativity of Early Childhood Education students using multi-source learning data analysis and unsupervised machine learning. The framework integrates six data categories comprising 30 features, utilizing the K-Means clustering algorithm (an unsupervised learning method) and a Creativity Index (CI) to facilitate result interpretation. Two independent simulated data scenarios - each containing 360 profiles - were used to test the process’s feasibility, reproducibility, and stability. In Scenario A, AI-MSC successfully recovered a pre-established three-group structure; K = 3 was consistently selected, and clustering results remained stable across initializations. In Scenario B, which featured a continuous data structure, K = 2 was only marginally preferred over K = 3, and the optimal number of clusters shifted when the weights of the assessment metrics were adjusted. The findings indicate that while AI-MSC is a feasible analytical framework for helping educators identify and interpret student profiles, it is not yet a validated tool for measuring creative competence. Further research is required to validate the framework using real student data and to assess the measurement validity of the feature system.