基于公众情感与体验的历史文化古镇保护研究——以苏州木渎古镇为例

Research on the Protection of Historical and Cultural Ancient Towns Based on Public Emotions andExperiences: Taking Suzhou Mudu Ancient Town as an Example

关键词:  风景园林  LDA主题模型  遗产保护  历史文化村镇  情感标绘

Key words:  landscape architecture  LDA theme model  heritage protection  historical and cultural village and town  emotional mapping

摘要:

公众对遗产的感知与认识不仅要考虑物质世界，还要考虑历史记忆、情感偏好和过程体验，公众对文化遗产和景观环境的感知过程与感知机制对于遗产 保护与活化极为重要。以苏州木渎古镇为例，基于LDA主题模型，结合情感四象限分析与情感标绘，挖掘与文化遗产相关的地方记忆和情感反馈，结合GIS搭建 数字化平台，实现情感数据可视化，从而揭示公众心理偏好，挖掘其背后的影响因素，并提出保护策略，以期为历史文化古镇的遗产保护与开发提供参考。

Abstract:

This study is situated within the digital era context, employing Mudu Ancient Town in Suzhou as a representative case. By integrating computational social science methodologies with spatial analysis techniques, it systematically investigates the characteristics of public emotional experiences in historical and cultural ancient towns and their spatial differentiation patterns, aiming to provide data-driven decision-making support for heritage conservation. The research addresses three pivotal questions: 1) how to quantitatively characterize the diverse emotional experiences of the public toward cultural heritage; 2) what mechanisms link different emotional types to heritage elements; and 3) how to formulate differentiated conservation strategies based on emotional analysis results. These questions respond to practical challenges in contemporary heritage conservation practices, such as inadequate public participation and insufficient evidence-based management. The study selects Mudu Ancient Town as its research subject due to its status as a quintessential Jiangnan water town facing representative challenges in tourism development and heritage conservation. The research adopts a multi-source data integration framework. During data collection, geotagged reviews were extracted from Sina Weibo. For sentiment analysis, an enhanced SnowNLP algorithm was employed to calculate sentiment values. Emotions were classified using a four-quadrant model: high pleasure (≥0.8), low pleasure (0.6-0.8), low displeasure (0.2-0.4), and high displeasure (＜0.2). In the topic modeling phase, the Latent Dirichlet Allocation (LDA) algorithm was applied to extract latent topics from textual data. Through topic coherence score computation and expert evaluation, 11 thematic factors were ultimately identified. The analysis yielded several significant findings. First, emotional experiences exhibit distinct spatial differentiation. Highpleasure emotions (sentiment value≥0.8) form prominent hotspots at core heritage sites such as Yan's Garden and Hongyinshanfang. Semantic network analysis indicates these emotions primarily derive from architectural artistry's visual impact and participatory engagement in cultural activities. Low-pleasure emotions (0.6-0.8) predominantly occur in natural landscape areas, including the Xiangxi River banks and Lingyan Mountain trails, with lexical analysis revealing terms like "tranquil" and "therapeutic", reflecting these environments' psychological restorative functions. Negative emotion analysis demonstrated that high-displeasure clusters (＜0.2) concentrate near the Qizi Mountain landfill (minimum sentiment value 0.15). Management issues at Shantang Old Street (e.g., ticket pricing, service quality) also constitute significant negative emotion sources. Low-displeasure (0.2-0.4) correlates with insufficient participatory experiences. The study further identified dynamic evolutionary characteristics in emotional experiences. First-time visitors exhibit higher high-pleasure emotion propensity (67.3%), while repeat visitors (≥3 visits) show increased low-pleasure experience tendency (58.1%), reflecting a demand evolution from "sightseeing novelty-seeking" to "in-depth experiential engagement". Based on these findings, the study proposes differentiated conservation strategies. For highpleasure areas, immersive cultural experience projects are recommended, such as recreating Emperor Qianlong's southern tour scenarios or developing ARguided systems to enhance interactive experiences and prolong engagement. For low-pleasure areas, ecological therapeutic function integration is suggested, including forest bathing trails and meditation spaces. For negative emotion concentration areas, management optimization solutions are proposed: dynamic pricing systems, service quality evaluation mechanisms, and waste management improvements. Notably, the study emphasizes community participation, advocating heritage conservation associations and community co-management applications to enhance resident involvement. This study represents the inaugural integration of LDA topic modeling, sentiment computation, and GIS spatial analysis, constructing a cultural heritage emotional analysis framework. The proposed "value-emotion" spatial database visually correlates heritage value with public experience, offering innovative conservation planning tools. The research outcomes possess transferability beyond ancient towns to diverse cultural heritage types, significantly advancing heritage management's scientific precision. As digital technology evolves, data-driven emotional research will assume increasingly pivotal roles in conservation decision-making.