Scholarlink

<p>Under post-productivist rural transformation, China’s rural revitalization has increasingly relied on a culture–tourism integration model that often marginalizes residents’ cultural identity and undermines long-term sustainability. As spaces where economic exchange, social interaction, and cultural memory converge, rural marketplaces constitute a distinctive yet underexplored type of heritage space. Using the Weichuan Marketplace in Henan, China, as a case study, we examine how residents’ perceptions of cultural landscape genes shape place attachment and discuss the implications for the conservation and development of comparable heritage spaces elsewhere. Based on fieldwork and documentary verification, we developed a cultural landscape gene framework for rural marketplaces and analyzed 823 valid questionnaires using an Optuna-optimized XGBoost-SHAP interpretable machine-learning model, which achieved an <i>R</i><sup>2</sup> of 0.884. The results show three main findings. First, the cultural landscape of rural marketplaces functions as a form of living heritage rooted in everyday practice. Second, at the element level, local delicacies (0.1533), neighborhood interaction (0.1269), and traditional craftsmanship (0.1007) made the strongest contributions; at the category level, cultural custom genes dominated (36.67%), highlighting the marketplace’s role in cultural transmission; and at the dimensional level, material (0.4378) and spiritual (0.3902) cultural genes contributed in a broadly balanced manner. Third, perceptions of cultural genes and place attachment exhibited a complex nonlinear relationship characterized by threshold and saturation effects. These findings extend the application of cultural landscape gene theory, demonstrate the value of interpretable machine learning for capturing complex human–place relationships, and support a strategy of residents’ core use with tourists’ compatible participation to promote the sustainable development of rural marketplaces and related heritage spaces.</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

The influence of cultural landscape gene perception on residents’ place attachment in rural marketplaces: an interpretable machine learning analysis based on XGBoost–SHAP

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月29日

查看全文
  • Mingyuan Liu,
  • Junping Xu,
  • Xu Kong,
  • Yixuan Shi

摘要

Under post-productivist rural transformation, China’s rural revitalization has increasingly relied on a culture–tourism integration model that often marginalizes residents’ cultural identity and undermines long-term sustainability. As spaces where economic exchange, social interaction, and cultural memory converge, rural marketplaces constitute a distinctive yet underexplored type of heritage space. Using the Weichuan Marketplace in Henan, China, as a case study, we examine how residents’ perceptions of cultural landscape genes shape place attachment and discuss the implications for the conservation and development of comparable heritage spaces elsewhere. Based on fieldwork and documentary verification, we developed a cultural landscape gene framework for rural marketplaces and analyzed 823 valid questionnaires using an Optuna-optimized XGBoost-SHAP interpretable machine-learning model, which achieved an R2 of 0.884. The results show three main findings. First, the cultural landscape of rural marketplaces functions as a form of living heritage rooted in everyday practice. Second, at the element level, local delicacies (0.1533), neighborhood interaction (0.1269), and traditional craftsmanship (0.1007) made the strongest contributions; at the category level, cultural custom genes dominated (36.67%), highlighting the marketplace’s role in cultural transmission; and at the dimensional level, material (0.4378) and spiritual (0.3902) cultural genes contributed in a broadly balanced manner. Third, perceptions of cultural genes and place attachment exhibited a complex nonlinear relationship characterized by threshold and saturation effects. These findings extend the application of cultural landscape gene theory, demonstrate the value of interpretable machine learning for capturing complex human–place relationships, and support a strategy of residents’ core use with tourists’ compatible participation to promote the sustainable development of rural marketplaces and related heritage spaces.