Capturing Perceived Characteristics of the Built Environment Using Artificial Intelligence and Urban Big Data

다국어 초록 (Multilingual Abstract)

Urban expansion is accelerating worldwide, with two-thirds of the global population projected to live in urban areas by 2050, implying that the urban built environment will affect the majority of humanity. This rapid transformation has profound implications for mental and emotional well-being, as the prioritization of economic growth over sustainability has degraded urban quality, creating dense, chaotic environments that heighten stress, anxiety, and other psychological health risks. Traditional urban analytics tools struggle to address the complexity of modern cities, limiting planners' ability to design responsive and inclusive urban spaces. Insufficient scientific frameworks to assess how the built environment influences mental and emotional health further exacerbate this challenge, often resulting in designs that neglect the psychological needs of urban dwellers.
Against this backdrop, this dissertation investigates the perceived characteristics of the built environment and their impact on psychological well-being in urban contexts. It addresses two primary questions: (1) How do individuals perceive various aspects of urban environments? and (2) What is the relationship between these perceptions and mental health outcomes? Employing a mixed-methods approach, the study integrates quantitative urban big data analysis with qualitative community survey insights to enhance understanding of the interaction between urban environments and psychological experiences. Leveraging advances in urban big data and artificial intelligence (AI), this work introduces an AI-based framework emphasizing mental and emotional responses, including perceived stress and place perception. Key contributions include:
1. ML-Based Stress Assessment: A machine learning approach for assessing perceived stress imposed by macro and micro urban features, utilizing open-source data to support planning in data-scarce regions like developing countries.
2. Enhanced Place Perception Framework: A conceptual framework integrating multisensory data and transfer learning to address visual bias and limited model generalizability in existing practices.
3. Exploration of Spatial Correlations: The first known analysis of spatial correlations between positive and negative place perceptions and perceived stress, revealing the intricate interplay between the built environment and psychological responses.
This research provides actionable insights for urban planners, policymakers, and environmental psychologists, enabling them to better evaluate how the built environment shapes psychological experiences. By identifying critical gaps in current practices and advocating for data-driven, community-focused strategies, the study underscores the need for a paradigm shift in urban planning that prioritizes psychological well-being and enhances urban quality of life.

목차 (Table of Contents)

• List of Figures vi
• List of Tables viii
• Abstract ix
• Chapter 1 Introduction 1
• 1.1. Research background 1
• 1.2. Context and motivation 3
• 1.3. Research objectives 7
• 1.4. Research Significance 9
• 1.4.1. Theoretical contributions 9
• 1.4.2. Practical contributions 10
• 1.5. Dissertation structure 10
• Chapter 2 Theoretical Considerations 12
• 2.1. Introduction 12
• 2.1.1. The built environment 13
• 2.1.2. Psychological well-being 14
• 2.1.3. Perceived characteristics of the built environment 15
• 2.2. Mental and emotional response 19
• 2.2.1. Perceived Stress 20
• 2.2.2. Place Perceptions 26
• 2.3. New trajectory in capturing perceived characteristics of the built environment 33
• 2.3.1. Advancing Urban Informatics and Data Science for enhancing
• urban planning and management 33
• 2.3.2. Novel approaches in perceived characteristics assessment 36
• 2.4. Summarize 37
• Chapter 3 Research Design and Methodology 39
• 3.1. Introduction 39
• 3.2. Theoretical framework and methodology 40
• 3.3. Material and sampling 43
• 3.3.1. Study location 43
• 3.3.2. Data acquisition and sampling 44
• Chapter 4 Measuring Perceived Stress—Mental response 48
• 4.1. Introduction 48
• 4.2. AI and Space Syntax theory in assessing Perceived Stress 49
• 4.3. Perceived Stress attribute selection 50
• 4.4. Model development process 52
• 4.4.1. Data acquiring and preprocessing 53
• 4.4.2. Characteristics of Seoul built environment 57
• 4.4.3. Model training and deployment 58
• 4.5. Results 61
• 4.5.1. Model performance and validation 61
• 4.5.2. Mapping Perceived stress 63
• 4.5.3. Cross-correlation between perceived stress indicators. 69
• 4.6. Discussion 71
• 4.7. Conclusion 75
• Chapter 5 Capturing Place Perceptions—Emotional responses 77
• 5.1. Introduction 77
• 5.2. The feasibility of utilizing regenerated sound in place perceptions assessment 78
• 5.2.1. Sound and Place Perceptions 79
• 5.2.2. Regenerated sound in place perceptions assessment 80
• 5.2.3. Visual-aural input generation 83
• 5.2.4. Feasibility testing under different conditions 85
• 5.2.5. Evidence of feasibility 88
• 5.3. Developing scalable predictive model for large-scale assessment 99
• 5.3.1. Model development process 100
• 5.4. Results 106
• 5.4.1. Model performance 106
• 5.4.2. Accuracy validation 108
• 5.4.3. Capturing perception of place 111
• 5.4.4. Correlation between place perception indicators 115
• 5.5. Discussion 118
• 5.6. Conclusions 123
• Chapter 6 Spatial Relationship of Perceived Characteristics 125
• 6.1. Introduction 125
• 6.2. Relationships of mental and emotional responses 127
• 6.3. Multiscale Geographically Weighted Regression (MGWR) 128
• 6.3.1. MGWR Model 130
• 6.3.2. Validate MGWR Model 131
• 6.4. Results interpretation and discussion 136
• 6.4.1. Output features 136
• 6.4.2. Global and Local policy—insight from MGWR analysis 142
• 6.4.3. Hypothesis testing 145
• 6.5. Conclusions 146
• Chapter 7 Conclusions and Future Works 149
• 7.1. Summarize and Conclusions 149
• 7.1.1. Measuring Perceived Stress 150
• 7.1.2. Capturing Place Perceptions 151
• 7.1.3. Spatial correlation of mental and emotional responses 152
• 7.2. The practical contribution and broader impacts 153
• 7.3. Recommendations for future research 155
• 7.3.1. Psycho-social sustainability 155
• 7.3.2. Development of psycho-comfort metric 157
• 7.4. References 159
• Abstract in Korean 189
• Acknowledgement 191