Scholarlink

<p>With the development of generative artificial intelligence (GenAI), image generation and text processing have drawn significant attention. However, thematic interpretation, particularly in urban design, remains unexplored. Therefore, this study investigates the potential of GenAI for thematic analysis in urban design. Interview data (<i>n</i> = 25) were analyzed using three GenAI platforms: ChatGPT, DeepSeek, and Gemini. First, each platform performed thematic coding to identify potential themes. Second, expert evaluations were compared with GenAI outputs to assess validity. Third, consistency was evaluated using Cohen’s <i>κ</i>. The results revealed nine place factors for the sense of place. While the consistency across GenAI tools was low, with –&#xa0;0.353, –&#xa0;0.419, and 0.528, respectively, their outputs collectively matched expert results. This suggests that GenAI is feasible for conducting thematic analysis, although variations exist. The findings provide a methodological foundation for qualitative urban research with GenAI, supporting its further integration into urban design-related decision-making and practice.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Potential of GenAI for thematic analysis in urban design: a pilot study on the sense of place

  • 期刊论文

  • 发表日期：

    2025年10月18日

查看全文
  • Ziyi Han,
  • Nor Haslina Ja’afar,
  • Mohd Iskandar Abd Malek,
  • Maslina Jamil,
  • Yuyan Lyu

摘要

With the development of generative artificial intelligence (GenAI), image generation and text processing have drawn significant attention. However, thematic interpretation, particularly in urban design, remains unexplored. Therefore, this study investigates the potential of GenAI for thematic analysis in urban design. Interview data (n = 25) were analyzed using three GenAI platforms: ChatGPT, DeepSeek, and Gemini. First, each platform performed thematic coding to identify potential themes. Second, expert evaluations were compared with GenAI outputs to assess validity. Third, consistency was evaluated using Cohen’s κ. The results revealed nine place factors for the sense of place. While the consistency across GenAI tools was low, with – 0.353, – 0.419, and 0.528, respectively, their outputs collectively matched expert results. This suggests that GenAI is feasible for conducting thematic analysis, although variations exist. The findings provide a methodological foundation for qualitative urban research with GenAI, supporting its further integration into urban design-related decision-making and practice.