The spatio-temporal characteristics of tourist flow is one of the core topics in tourism geography. However, few studies have paid attention to community analysis in tourist flow network. This paper uses mobile signaling data based on random walk approach to identify the communities of tourist flow. A node influence evaluation is established to identify the core nodes of the community. Popularity response concept is introduced and factors influencing the popularity response of the community is analysed. The results indicate: 1) Spatially, it identifies 4 communities, extracting the differential rules of each community, and suggests the location proximality in formation of communities; 2) By comparing different staying period, it reveals the role of core node such as transport hubs, key attractions and emerging attractions in different staying period. This study also reveals that accommodation density, the significance of core node, key tourist attractions and accessibility have significant influence on the formation of communities. By adding a detailed time dimension, the paper identifies tourism communities and community core nodes and analyses the important factors influencing their popularity response, which contributes to the formation of potential tourism community, the marketing of tourism community, and tourism activities prediction.

旅游流时空特征是旅游地理学的核心课题之一。然而，很少有研究关注旅游流网络中的社区分析。本文利用基于随机游走方法的移动信令数据来识别旅游流的社区。建立节点影响力评价，识别社区核心节点。介绍了人气响应的概念，分析了影响社区人气响应的因素。结果表明：1）在空间上，识别出4个社区，提取每个社区的差异规则，并建议社区形成的位置邻近性；2）通过比较不同停留时间，揭示交通枢纽等核心节点的作用，不同停留时段的重点景点和新兴景点。研究还揭示了住宿密度、核心节点重要性、重点旅游景点及可达性对社区形成具有显着影响。通过添加详细的时间维度，识别旅游社区和社区核心节点，分析影响其人气反应的重要因素，有助于潜在旅游社区的形成、旅游社区的营销、旅游活动的预测。