面向AI应用的青年一人居住空间感性体验与CMF偏好的中介效应研究

A Mediation Analysis of Emotional Experience and CMF Preferences in Youth Single-Person Living Spaces for AI Applications

摘要: 近年来，中国城市青年一人居住现象日益普遍，相应的空间感性需求与生活方式差异已成为居住设计的重要议题。与此同时，生成式AI正在改变空间设计的概念生成与可视化方式，但生成质量仍高度依赖用户能否以具体且结构化的语言表达需求。基于此，本研究以中国城市地区18~35岁的青年一人居住者为对象，对355份有效样本进行分析，考察其居住空间感性体验、功能空间认知与CMF偏好之间的关系，并为后续AI提示语结构化提供依据。研究结果表明：当前居住空间中，支配感与愉悦感相对较高，唤醒感相对较低；部分功能空间在秩序感与尺度感方面存在显著的现状–理想差异；多数功能空间对色彩维度的偏好较高，休息空间则对材料维度评价更高。进一步分析显示，感性体验、空间认知与CMF偏好之间存在显著正向关联，理想空间认知需求在感性体验与CMF偏好之间呈现中介关联特征。需要指出的是，本研究基于横断面数据，相关结果应理解为变量之间的关联结构与可能路径，而不宜作严格因果推断。研究从结构层面揭示了青年一人居住者的感性需求及其与CMF偏好之间的联系，可为生成式AI住宅空间设计中的提示语组织与感性导向设计策略提供参考。

Abstract: Single-person living among urban youth in China has become increasingly common, making emotional needs and lifestyle diversity in residential space an important design issue. At the same time, generative AI is reshaping concept development and visualization in spatial design, yet the quality of generated outcomes still depends heavily on whether users can express their needs in concrete and structured language. Against this background, this study examined 355 valid responses from urban Chinese young adults aged 18~35 who live alone and explored the relationships among emotional experience in living space, spatial cognition of functional areas, and CMF preferences, with the aim of informing prompt structuring for later AI applications. The results show that dominance and pleasure were relatively high in current living spaces, whereas arousal was relatively low. Significant current-ideal differences appeared mainly in orderliness and scale in some functional spaces. Across most functional areas, respondents showed stronger preferences for the color dimension, while the material dimension was rated higher in rest spaces. Further analyses indicate significant positive associations among emotional experience, spatial cognition, and CMF preferences, and ideal spatial cognition showed a mediation-like association between emotional experience and CMF preferences. Because the study is based on cross-sectional data, these findings should be interpreted as associations, predictive relationships, and possible pathways rather than strict causal effects. Overall, the study reveals the structural links between emotional needs and CMF preferences in single-person living spaces and provides empirical support for structured prompt design in AI-assisted residential spatial design.