From Context to Prompt: CSPM-Based Structured Prompting in AI-Assisted Spatial Design

Abstract

This study proposes the Context – Schematic – Perceptual – Meaning (CSPM) framework as a prompt-engineering method for AI-assisted design. As a schematic prompting approach, CSPM helps organize the interior design brief and its structure, including multisensory design and cultural contexts, into a structurally interpretable prompt that aligns with the design intention to generate outcomes. These structured parameters guide more intentional and logical design generation by limiting excessive semantic drift and iterative dependencies, which are common challenges in human-AI interaction. To examine operational use, a comparative case study was conducted using questionnaires during a small-group practitioner reflection. The survey questionnaires were designed to compare visual design results between the unstructured and CSPM-structured prompts, using the same case study. The results show that an unstructured prompt may produce visually sophisticated outputs but exhibit more iterative stages, indicate some semantic drift, and have weak traceability to encoded cultural values. Meanwhile, the CSPM results suggest that it may reduce iteration dependence and provide clearer alignment with design logic intent. The CSPM reframes prompting as a human-centered AI orientation, while maintaining a systemic approach to prompt engineering to achieve interpretability, reduce ambiguity, and ensure fidelity of intent, thereby generating generative workflows.