Scholarlink

<p>Cultural routes, once valued as heritage corridors, are increasingly experienced as tourism destinations composed of multiple attractions. Unlike work centered on tourists’ subjective motivations, this study examines how objective destination attributes shape tourist visitation along routes. Using the Kumano Kodo (Japan), 24,569 geo-tagged Flickr records quantify visitation patterns. Topic modeling of textual descriptions in the same records identifies 17 destination attributes, grouped into 4 categories. Multiple data-processing methods transformed geographic information into machine-learning-ready datasets, which are evaluated across several models. The best configuration is interpreted with PDP, heatmaps, and SHAP. Results indicate that visitation arises from combined, nonlinear attribute effects: cultural and heritage attractions function as primary anchors, tourism / commercial facilities and transport conditions provide essential support, while natural environments mainly moderate context. Attribute combinations also vary across seasons, travel modes, and visitor preferences. These findings offer practical insights and provide a transferable framework for understanding cultural-route tourism mechanisms.</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

Understanding how destination attributes shaping tourist visitation on cultural routes through social media data and interpretable machine learning

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月30日

查看全文
  • Xinyue Lin,
  • Xiao Teng,
  • Zhenjiang Shen,
  • Qizhi Mao

摘要

Cultural routes, once valued as heritage corridors, are increasingly experienced as tourism destinations composed of multiple attractions. Unlike work centered on tourists’ subjective motivations, this study examines how objective destination attributes shape tourist visitation along routes. Using the Kumano Kodo (Japan), 24,569 geo-tagged Flickr records quantify visitation patterns. Topic modeling of textual descriptions in the same records identifies 17 destination attributes, grouped into 4 categories. Multiple data-processing methods transformed geographic information into machine-learning-ready datasets, which are evaluated across several models. The best configuration is interpreted with PDP, heatmaps, and SHAP. Results indicate that visitation arises from combined, nonlinear attribute effects: cultural and heritage attractions function as primary anchors, tourism / commercial facilities and transport conditions provide essential support, while natural environments mainly moderate context. Attribute combinations also vary across seasons, travel modes, and visitor preferences. These findings offer practical insights and provide a transferable framework for understanding cultural-route tourism mechanisms.