Using Novel Multi‐Piece Tangram as a Manipulating Tool for STEAM Curriculum Design: An Optimization‐Based Experimental Approach to Aesthetic Education

Abstract

It has also been noted that the literature is scarce on the inclusion of aesthetic education under the STEAM framework. To address this gap, this study proposes an optimized model titled T‐STEM‐ICA‐SVMAP, which includes the integration of Support Vector Machine based Aesthetic Perception (SVMAP) and the Imperialist Competitive Algorithm (ICA). This study aims to improve the cognitive process of learning and aesthetic perception under the STEAM framework. This study has utilized an experimental study design with four specific research questions: optimization methods, identification of aesthetic components, comparative study of the proposed model, and evaluation of the reproducibility of the system. This study has utilized the publicly available Tangram image dataset available on the Kaggle platform, which includes diverse representations of tangram images. An 80:20 train‐test split is applied. Features extracted with SVMAP for visual and geometry, while dynamic curriculum parameters of ICA for the difficulty of the task, geometry restrictions, and activity sequence. The framework was compared with two baseline models: a traditional SVM and a CNN‐based classifier. The results show that the proposed framework achieved an accuracy of 0.89, precision of 0.88, recall of 0.87, and F1‐score of 0.87, outperforming the two baselines by 0.05–0.11 in all metrics. The findings indicate that combining optimization‐driven curriculum design with machine learning‐based aesthetic evaluation provides an effective, objective, and reproducible means of integrating aesthetic education into tangram‐based STEAM learning. This study contributes a replicable computational‐ pedagogical pipeline to educational technology and STEAM curriculum research.