簡易檢索 / 詳目顯示
| 研究生： |
簡芷嫺 Chien, Chih-Hsien |
|---|---|
| 論文名稱： |
人物誌導向之AI代理人於自動化UX診斷之應用：以年長者電商使用情境為例 Persona-Aware AI Agents for Automated UX Diagnosis: A Case Study of Elderly Users in E-commerce Websites |
| 指導教授： |
簡士鎰
Chien, Shih-Yi |
| 口試委員: |
簡士鎰
Chien, Shih-Yi 林斯寅 Lin, Szu-Yin 康藝晃 Kang, Yi-Huang |
| 學位類別： |
碩士
Master |
| 系所名稱： |
商學院 - 資訊管理學系 Department of Management Information System |
| 論文出版年： | 2026 |
| 畢業學年度： | 114 |
| 語文別： | 英文 |
| 論文頁數： | 151 |
| 中文關鍵詞： | 大型語言模型 、人工智慧代理人 、自動化UX診斷 、使用者體驗 |
| 外文關鍵詞： | LLM, AI Agent, Automated UX Diagnosis, User Experience |
| 相關次數： | 點閱：47 下載：0 |
| 分享至: |
| 查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
隨著全球高齡化趨勢加劇，如何設計符合高齡者需求的數位服務已成為人機互動（Human–Computer Interaction, HCI）領域的重要議題。傳統使用者體驗（User Experience, UX）評估方法，如可用性測試，雖能有效發現網站可用性問題，但需投入大量人力、時間及成本，難以支援持續性的網站改善。近年來，大型語言模型（Large Language Models, LLMs）與 AI Agent 的發展，使 AI 輔助 UX 評估成為可能。然而，現有 AI Agent 多以任務完成為導向，較難深入理解高齡者的使用者體驗，因此仍難以有效支援高齡者 UX 評估與網站持續優化。本研究提出 Agent-driven UX Optimization Framework，整合高齡者人物誌（Persona）、具反思能力之雙迴圈（Dual-loop）AI Agent、UX Analyzer 與持續性網站改善流程，探討反思思考機制是否能提升 AI Agent 對高齡者 UX 問題的辨識與成因分析能力，並支援網站持續優化。研究比較 Single-loop 與 Dual-loop AI Agent 在相同電商購物任務下的 UX 評估結果，以高齡者可用性測試作為驗證基準，並透過 UX Analyzer 產生設計建議，完成網站改善與再次評估。研究結果顯示，Dual-loop AI Agent 較能辨識符合高齡者特性的 UX 問題，並透過 Reflection 與 Wonder 提供更深入的問題成因分析。UX Analyzer 能有效將 UX 問題轉換為具體設計建議，經網站改善與重新測試後，主要 UX 問題皆獲得改善，驗證本研究所提出 AI 輔助 UX 優化流程的有效性。本研究主要貢獻包括：（1）驗證反思思考機制可提升 AI Agent 對高齡者 UX 問題之辨識與成因分析能力；（2）提出具可解釋性的 UX Analyzer，以支援 UX 診斷與設計建議生成；（3）建立整合 AI Agent、UX 診斷、網站改善與再評估之 Agent-driven UX Optimization Framework，為 AI 輔助 UX 評估與高齡友善數位服務設計提供新的研究方向。
As global population aging accelerates, designing digital services that meet the needs of older adults has become an important topic in Human–Computer Interaction (HCI). Traditional user experience (UX) evaluation methods, such as usability testing, effectively identify usability issues but require substantial time and cost, limiting continuous website improvement. Recent advances in Large Language Models (LLMs) and AI agents have enabled AI-assisted UX evaluation. However, most existing AI agents are task-oriented and have limited ability to understand the interaction experiences of older adults. This study proposes an Agent-driven UX Optimization Framework integrating elderly personas, a reflection-enabled Dual-loop AI Agent, a UX Analyzer, and a continuous website improvement process. The framework investigates whether reflective reasoning enhances AI agents' ability to identify elderly-related UX issues, analyze their underlying causes, and support website optimization. Single-loop and Dual-loop AI agents are evaluated using the same e-commerce shopping task, with usability testing involving older adults serving as the validation benchmark. Based on the evaluation results, the UX Analyzer generates design recommendations for website improvement and re-evaluation. The results show that the Dual-loop AI Agent more effectively identifies elderly-related UX issues and provides deeper analyses of their underlying causes through the Reflection and Wonder mechanisms. After implementing the generated recommendations, the major UX issues were substantially improved, demonstrating the effectiveness of the proposed framework. This study makes three contributions: (1) verifying that reflective reasoning enhances AI agents' ability to identify elderly-related UX issues and analyze their underlying causes; (2) proposing an interpretable UX Analyzer for UX diagnosis and design recommendation generation; and (3) establishing an Agent-driven UX Optimization Framework for AI-assisted UX evaluation and the design of age-friendly digital services.
摘要 2
Abstract 3
Table of Contents 4
Tables 9
Figures 10
Chapter 1 INTRODUCTION 11
1.1 Research Background 11
1.2 Research Motivation 12
1.3 Research Gap 14
1.4 Research Objectives 15
1.5 Research Questions 17
1.6 Research Contributions 20
Chapter 2 Literature Review 22
2.1 Usability and User Experience (UX) Evaluation 22
2.2 Usability Challenges for Older Users 24
2.3 AI Agents for UI Interaction 25
2.4 Persona-Based UX Analysis 27
2.5 Single-loop and Dual-loop Agent Architecture 28
2.6 AI-assisted UX Evaluation and Research Positioning 30
Chapter 3: Methodology 32
3.1 Research Framework 32
3.2 Research Materials 37
3.2.1 Experimental Website 37
3.2.2 Task Scenario 39
3.2.3 Data Sources 40
3.3 Elderly Persona Construction 41
3.3.1 Visual Characteristics 43
3.3.2 Motor Characteristics 43
3.3.3 Cognitive Characteristics 44
3.3.4 Affective Characteristics 44
3.3.5 Persona Integration into AI Agents 45
3.4 Single-loop Agent Design 45
3.4.1 Single-loop Agent Architecture 46
3.4.2 Environmental Perception 47
3.4.3 Task-oriented Decision Making 48
3.4.4 Absence of Reflective Reasoning 48
3.4.5 Role in This Study 49
3.5 Dual-loop Agent Design 50
3.5.1 Dual-loop Architecture 51
3.5.2 Fast Loop 52
3.5.3 Slow Loop 53
3.5.4 Reflection Mechanism 54
3.5.5 Wonder Mechanism 55
3.5.6 Memory System 56
3.5.7 Role in This Study 57
3.6 UX Analyzer Design 59
3.6.1 Overview of the UX Analyzer 60
3.6.2 Layer 1: Data Preparation 61
3.6.3 Layer 2: UX Event Extraction 62
3.6.4 Layer 3: Guideline-based UX Reasoning 63
3.6.5 UX Analyzer Output 65
3.6.6 Role in This Study 66
3.7 Elderly UX Guideline Framework 66
3.8 Website Improvement Process 69
3.8.1 Overview of the Website Improvement Process 69
3.8.2 Website Version 1 71
3.8.3 Design Recommendation Generation 71
3.8.4 Website Version 2 72
3.8.5 Re-testing and Evaluation 73
3.8.6 Role in This Study 74
3.9 Experimental Design 75
3.9.1 Experimental Shopping Website 76
3.9.2 Experimental Environment 79
3.9.3 Participant Demographics 79
3.9.4 Experimental Phase 1: Elderly UX Diagnosis 81
3.9.5 Experimental Phase 2: Website Improvement and Re-evaluation 82
3.9.6 Research Question Mapping 82
3.9.7 Summary of Experimental Comparisons 84
Chapter 4: Results 85
4.1 UX Issues Identified from Older Adult Participants 85
4.2 RQ1-1: UX Pain Point Discovery 89
4.3 RQ1-2: Root Cause Analysis Through Reflection 92
4.4 RQ1-3: Human Similarity Analysis 96
4.5 RQ2-1: Design Recommendation Generation 99
4.6 RQ2-2: Website Improvement Effectiveness 102
4.7 Summary of Findings 106
Chapter 5: DISCUSSION 109
5.1 Discussion of Objective 1: Reflection-enhanced UX Pain Point Discovery 109
5.2 Discussion of Objective 2: Reflection-supported Root Cause Analysis 112
5.3 Discussion of Objective 3: Agent-driven UX Optimization Framework 115
5.4 Theoretical Contributions 118
5.4.1 Extending AI Agent Evaluation Beyond Task-oriented Performance 118
5.4.2 Reflection as a Mechanism for UX Diagnosis and Root Cause Analysis 119
5.4.3 Establishing an AI-assisted UX Optimization Framework 120
5.5 Practical Implications 121
5.5.1 Implications for UX Design Practice 121
5.5.2 Implications for Organizational UX Processes 122
5.5.3 Implications for AI-assisted UX Evaluation 123
5.5.4 Implications for Digital Inclusion of Older Adults 124
5.6 Limitations 125
5.6.1 Single Platform Evaluation 125
5.6.2 Single Task Scenario 126
5.6.3 Limited Older Adult Sample 126
5.6.4 Subjectivity in Human–Agent Mapping 126
5.6.5 Persona-based Simulation 127
5.6.6 Single Dual-loop Agent Architecture 127
5.6.7 Dependence on Large Language Models 127
5.7 Future Research Directions 128
5.7.1 Generalizing AI-assisted UX Evaluation Across Diverse Digital Contexts 128
5.7.2 Toward Personalized Persona-driven UX Evaluation 129
5.7.3 Advancing Cognitive AI Agents for UX Evaluation 129
5.7.4 Human–AI Collaborative UX Evaluation 130
5.7.5 Toward Fully Automated Agent-driven UX Optimization 131
Chapter 6: Conclusion 132
6.1 Research Summary 132
6.2 Major Findings 133
6.3 Final Conclusion 137
Reference 139
Appendix A. Complete Elderly UX Design Guideline Framework 141
Appendix B. Prompt Design of the UX Analyzer 149
Adlin, T., Pruitt, J. S., Goodwin, K., Hynes, C., McGrane, K., Rosenstein, A., & Muller, M. J. (2006). Putting personas to work. CHI '06 Extended Abstracts on Human Factors in Computing Systems.
Amouzadeh, E., Dianat, I., Faradmal, J., & Babamiri, M. (2025). Optimizing mobile app design for older adults: systematic review of age-friendly design. Aging Clinical and Experimental Research, 37.
Cooper, A. (1999). The Inmates are Running the Asylum. Software-Ergonomie.
Czaja, S. J., Boot, W. R., Charness, N., & Rogers, W. A. (2019). Designing for older adults: Principles and creative human factors approaches (3rd ed.). CRC Press.
Czaja, S. J., & Lee, C. C. (2007). Information technology and older adults.
Deng, X., Gu, Y., Zheng, B., Chen, S., Stevens, S., Wang, B., Sun, H., & Su, Y. (2023). Mind2Web: Towards a Generalist Agent for the Web. ArXiv, abs/2306.06070.
Fisk, A. D., Rogers, W. A., Charness, N., Czaja, S. J., & Sharit, J. (2004). Designing for Older Adults: Principles and Creative Human Factors Approaches.
Hou, G., Anicetus, U., & He, J. G. (2022). How to design font size for older adults: A systematic literature review with a mobile device. Frontiers in Psychology, 13.
International Organization for Standardization. (2018). ISO 9241-11: Ergonomics of human-system interaction—Part 11: Usability: Definitions and concepts.
Ivory, M. Y., & Hearst, M. A. (2001). The state of the art in automating usability evaluation of user interfaces. ACM Comput. Surv., 33, 470-516.
Lee, C., & Coughlin, J. F. (2015). PERSPECTIVE: Older Adults' Adoption of Technology: An Integrated Approach to Identifying Determinants and Barriers. Journal of Product Innovation Management, 32, 747-759.
Liu, B., & Wang, C. (2024). Elderly-Centric Chromatics: Unraveling the Color Preferences and Visual Needs of the Elderly in Smart APP Interfaces. International Journal of Human–Computer Interaction, 41, 3527-3536.
Lu, Y., Yao, B., Gu, H., Huang, J., Wang, J., Li, Y., Gesi, J., He, Q., Li, T. J., & Wang, D. (2025). UXAgent: A System for Simulating Usability Testing of Web Design with LLM Agents. ArXiv, abs/2504.09407.
Mattick, X., Dogangün, A., Boadu, J., Jansen, N., & Meske, C. (2024). Design for Older People: Improving the Usability of Mobile Apps through Targeted Design Recommendations. Proceedings of the 36th Australasian Conference on Human-Computer Interaction.
Nielsen, J., & Molich, R. (1990). Heuristic evaluation of user interfaces. International Conference on Human Factors in Computing Systems.
Nielsen, P. L. (2019). Personas - User Focused Design. Human–Computer Interaction Series.
Ning, L., Liang, Z., Jiang, Z., Qu, H., Ding, Y., Fan, W., Wei, X., Lin, S., Liu, H., Yu, P. S., & Li, Q. (2025). A Survey of WebAgents: Towards Next-Generation AI Agents for Web Automation with Large Foundation Models. Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2.
Park, J. S., O’Brien, J. C., Cai, C. J., Morris, M. R., Liang, P., & Bernstein, M. S. (2023). Generative Agents: Interactive Simulacra of Human Behavior. Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology.
Wharton, C., Rieman, J., Lewis, C. H., & Polson, P. G. (1994). The cognitive walkthrough method: a practitioner's guide.
World Health Organization. (2021). Working together to promote digital inclusion for older people. https://www.decadeofhealthyageing.org/find-knowledge/resources/publications/working-together-to-promote-digital-inclusion-for-older-people
