Research on the Influencing Mechanism of Historic Urban Landscape Characteristics on Public Sentiments and the Spatio-temporal Differentiation Patterns—A Case Study of Shaoxing Ancient City in Zhejiang Province, China

Abstract

Combining research methods such as semantic analysis and Space-Time Cube, this paper proposes a "historic urban landscape–cognition–sentiment" analysis framework, covering aspects of heritage intrinsic value, urban functional value, and urban landscape value, and reveals the influencing mechanism of the characteristics of historic urban landscape (HUL) on public sentiments and the spatio-temporal differentiation patterns through the case study of the Shaoxing ancient city in China. The research findings include that different HUL characteristics had differentiated effects on public sentiments, and the effect of a same HUL characteristic on public sentiments varied between weekdays and weekends/holidays, and among different HULs. On weekends/holidays, public sentiments were more influenced by the intrinsic value factors of HUL (e.g., heritage level, heritage age), whereas on weekdays, they were more affected by urban functional value factors, and urban landscape value factors played a greater role in arousing people's positive sentiments. This study aims to provide scientific references for enhancing public perception and emotional experience in urban spaces and for identifying potential spatial improvement opportunities in historic cities.

● Innovatively proposes "HUL–Cognition–Sentiment" analysis framework, and systematically explores the influencing mechanism of HUL characteristics on public sentiments

● Studies the spatio-temporal differentiation patterns of public sentiments from the dimensions of HUL intrinsic value, urban functional value, and urban landscape value

● On weekends/holidays, public sentiments were more influenced by the intrinsic value factors of HUL, while on weekdays were more affected by urban functional value and urban landscape value factors

Graphical abstract

Keywords

Historic Urban Landscape / Public Sentiments / Weibo Check-in Data / Spatio-temporal Differentiation Patterns / Space-Time Cube / Historic Urban Landscape–Cognition–Sentiment Analysis Framework

1 Introduction

In 2011, UNESCO issued The UNESCO Recommendation on the Historic Urban Landscape ("The Recommendation" hereafter), introducing the concept of "historic urban landscape" (HUL). HUL is defined as "the urban context and its geographical setting taking into consideration the historical layering of cultural and natural values and attributes"^[1]. This definition emphasizes the historical layering of cultural and natural legacies, as well as the dynamic superimposition of current urban development, where culture is seen as the driving force of the city, nature as its carrier, and urban landscape as the result of such interactions and evolution^[1]. The Recommendation also highlights the need to research the public's collective memory and universal consciousness in the preservation of HUL^[2]. The cognitive theory of emotion posits that emotions are products of the interactions between people and the environment^[3]. Recent studies have gradually unveiled the complexity of public sentiment perception, which is closely related to the attributes of HUL and the image of heritage sites^[4].

It is noteworthy that ancient towns or historic cities, as an important subclass of HUL, have garnered increasing attention in recent years^[5][6]. These cities and towns enjoy rich historical and cultural assets but also face various challenges brought by globalization and modernization, especially the exacerbated cultural homogenization and identity loss of HUL. The root cause of this issue lies in an overemphasis on production efficiency under short-sighted urban development ideologies, which neglect human needs and user experience in sustainable urban development^[7].

In recent years, public perception and emotional experience of physical environments have become a focal point in urban studies^[8], which, however, is less combined with HUL in academic efforts. Existing scholarly research predominantly focuses on public perception on the image of HUL, with subjects such as visual image perception of HUL based on digital footprints^[9], evaluation of HUL based on online reviews^[10], and heritage identity perception^[11]. In these studies, public sentiments and feelings are merely considered as indicators or representations of HUL perception, and their relationship with HUL preservation and sustainable development has not been fully explored. Therefore, this study proposes an "HUL–Cognition–Sentiment" framework that analyzes the influencing mechanisms of HUL characteristics of historic cities on public sentiments, hoping provide scientific references for enhancing public well-being and identifying potential improvement opportunities for historic cities.

2 Literature Review on Public Sentiments From the Perspective of HUL

2.1 Public Cognition and Sentiments From the Perspective of HUL

HUL not only encompasses physical spaces but also serves as a vessel for cultural and public historical memories, providing a profound and significant context for contemporary human life^[12]. HUL also involves intangible values of historic cities. Social values, community identity, and civic pride are not only reflected in monuments and buildings, but also embodied in the places and moments of people's collective life^[13]. From the perspective of HUL, people can form deep emotional connections with the places they live in^[14]. Rossana Bonadei et al. noted that when the public interact with HUL, they would have their own perception based on their aesthetic, emotional, cultural, and relational values or the experiences attached to it^[15]. For example, historians might emphasize the origins of HUL, while architects often highlight its artistic value and the use of specific materials. In this context, understanding public sentiments about HUL and how the perception influences their emotional experience becomes critical.

Since the formal introduction of Emotional Geography in 2001^[16], emotions have shifted from being purely subjective mental issues to broader socio-cultural fields, generating spatial, open, and relational emotions^[17]. Public emotion, as people's collective feelings and attitudes towards a public matter or event, is "often the product of repeated place interactions and experience"^[18]; while, perception can reveal the extent of public cognition regarding the given public matter or event. Richard Stanley Lazarus's cognitive theory of emotion further underscores the relationship between cognition and emotion, positing that "emotions are the products of appraisals"^[19]. This implies that the public's cognitive assessment of HUL, influenced by sensory reception and personal experience, directly affects their emotions. In summary, cognition is the basis of sentiments, and sentiments are an extension of cognition^[4]; the public's cognition and sentiments about HUL constitute a dynamic, interactive process. To better protect and utilize HUL, and to ensure that HUL planning and management align more closely with public needs and desires, it is essential to deeply understand the process of how the public's cognition and sentiments are generated concerning HUL.

2.2 Methods for Measuring Public Sentiments

With the rapid advance of location-based services (LBS) technology, it is able to actively or passively obtain and collect information about public sentiments and attitudes with sensing devices and positioning technologies, enabling the measurement and visualization of public sentiments within spatio-temporal units.

Typically, active measurement methods require the public's cooperative participation^[20]. For example, questionnaires^[21] can be used to collect in-depth insights into the public's emotional responses to a specific urban landscape; laboratory observations^[22] and wearable sensing devices^[23] can be used to capture real-time sentiment changes in urban spaces. These methods provide urban planners with direct and detailed feedback, allowing them to better understand the public needs and desires. In contrast, passive measurement methods do not require the public's active participation^[20]; aided by advanced big data semantic analysis technology, including various lexicons or machine learning algorithms^[24], social media data from platforms like X (former Twitter)^[25] and Weibo^[26] offer new opportunities for passive measurement and visualization of public sentiments. Such unstructured data, with user-posted spatial information, contain extensive textual information that can reflect citizens' real sentiments (e.g., happiness, sadness, fear, disgust, anger, surprise)^[27], helping reveal the spatio-temporal differentiation patterns of public sentiments. This approach has been widely applied in urban planning, for example, to study citizens' sentiments in response to public safety incidents^[28] and public sentiment changes towards the government following terrorist events^[29].

2.3 The Influence of HUL Characteristics on Public Sentiments

HUL is not a static built-up setting but the result of continuous dynamic layering, representing a complicated, continuously adaptive socio-ecological system^[30]. Sentiments are influenced by both activities and places, and the spatio-temporal sequences of activities and places are in turn influenced and constrained by emotions^[31]. This means that public sentiments of the same HUL can vary significantly across different times and spaces. Jia Jian et al. pointed out that the physical characteristics, spatial scale, and functional organization of the built-up setting would directly influence public sentiments and experiences^[31]. Zexin Lei et al. provided a theoretical framework, summarizing the criteria of HUL evaluation system into three aspects, namely protection status, intrinsic value, and integration with the city^[32].

At the aspect of intrinsic value, some studies have revealed how the intrinsic value of HUL influences public perception on historic cities, highlighting its core role in establishing emotional connections and fostering cultural identity. For example, a study using wearable devices to survey public emotional arousal in Jerusalem found that areas with the greatest protection value, such as the Temple Mount and the Wailing Wall, showed the highest levels of positive sentiments among the public^[33]. Additionally, research on public sentiments about landscapes with special values, such as dark heritage sites like Auschwitz, has also gained widespread attention in academia^[34].

However, research on the influence of the other two aspects, protection status and integration with the city, on public sentiments remains limited. First, the protection status of historical buildings and relics can influence people's identity and sense of belonging: well-preserved heritages may evoke the public's pride and honor, while neglected or damaged sites may lead to feelings of loss or anger^[35][36]. The specific mechanisms of these effects, however, are still underexplored. Second, the degree of HUL integration with the modern urban environment might affect the public's emotional experiences, which is similar to the influencing mechanisms of built environment that urban geography has long focused on. Zhuoran Shan et al., by collecting POI and Weibo data from the main urban areas of Wuhan, verified that the densities of transportation facilities, commercial outlets, jobs, entertainment facilities, public service facilities, and outdoor recreational spaces significantly affect residents' sentiments^[26]. Miguel Jesús Medina-Viruel et al. also confirmed that there is a significant positive correlation between entertainment and the accessibility and convenience of facilities with tourists' positive sentiments around the heritage sites of Úbeda and Baeza in Spain^[37]. Yali Yang et al. found that, for the public in Barcelona, sentiment-related Twitter posts tend to cluster around tourist attractions or recreational spots^[18].

In summary, although cities are considered dynamic and cumulative products of spatio-temporal changes, the influencing mechanisms of HUL characteristics on public sentiments remain insufficiently explored. Filling the gap, this study uses social media big data to measure public sentiments and to reveal the influencing mechanisms of which by HUL characteristics. This research aims to provide a reference for theoretical and practical efforts for identifying potential areas for improvement in historic cities and the development of optimization strategies in urban planning and management.

3 The HUL–Cognition–Sentiment Analysis Framework

To elucidate the dynamic evolution process from identifying HUL characteristics to the public's emotional responses, this study constructs an HUL–Cognition–Sentiment (HCS) analysis framework (Fig.1), which consists of three main parts.

Fig.1 HCS analysis framework.

Full size|PPT slide

1) Analysis of HUL characteristics. This study employs the Geographically Weighted Regression (GWR) model to reveal and examine the core characteristics of HUL in dimensions of heritage intrinsic value, urban functional value, and urban landscape value. Intrinsic value considers heritage grade and age to represent HUL preservation status and historical value. Urban functional value selects urban services and gained attention of heritages to represent the interactions between HUL and other functional structures in the city. Urban landscape value examines the openness of water bodies and green spaces and the degree of mixed land use to represent the visual and aesthetic integration of HUL with natural and urban environments.

2) Analysis of public sentiments. This study used Jieba Python segmentation module and Sentiment Lexicon to dissect textual big data from Weibo, combine individuals' direct feedback to HUL with deeper cognitive experiences, and empirically quantify the transformation process from sensory responses to emotional cognition—the public's sensory reception and experiences when identifying HUL characteristics generate sensory information, which is processed by the brain into cognition. In this research, the cognition results are expressed as positive, neutral or negative emotions.

3) Analysis of spatio-temporal distribution of public sentiments. Based on the measurements of public sentiments, a comparative study was conducted on the intensity and geographical distribution of sentiments during weekdays and weekends/holidays. Space-Time Cube and Getis-Ord Gi* index were used to reveal the spatio-temporal differentiation patterns of sentiments, exploring the changes and patterns of public sentiments. People's sentiments determine their active or passive feedback towards HUL, and in turn the assembly of these spatial behaviors dynamically influences the formation and evolution of HUL^[38]. There are two main paths for public feedback on HUL: direct emotional expression and spatial behavior without emotional expressions. For the former, it can be the public's favoritism or dissatisfaction with HUL expressed through platforms such as social media, which can directly influence HUL conservation and development strategies; while the latter, for instance, frequent visits to a particular HUL, can reflect the attractiveness of a given HUL to the public, which will in turn influence urban planning and cultural heritage management.

Combining these three parts, the HCS analysis framework provides a multi-dimensional perspective to explain the generation process of public sentiments about HUL. Demonstrating with the case of the Shaoxing ancient city in Zhejiang Province, China, this paper employs the HCS framework and Weibo big data, along with multiple spatial analysis methods, and explores the interactions between HUL and public sentiments.

4 Study Area and Research Methods

4.1 Study Area

The Shaoxing ancient city in Zhejiang Province was initially constructed in 490 BC, boasting a history of over 2, 500 years, as an area having evolved continuously through different historical periods^①. The study area is defined by the outer banks of the ring moat in the Yuecheng District of the city, covering an area of approximately 9 km^2[39]. The ancient city also enjoys unique water town landscapes characterized by "three mountains, hundreds of bridges, and thousands of winding alleys with interwoven waterways, " as well as a wealth of historical and cultural heritages. According to the Shaoxing Historical and Cultural City Protection Plan (2021–2035), the ancient city includes eight historical and cultural neighborhoods: Lu Xun's Former Residence, Yuezi Town, Bazi Bridge, the Former Residence of Calligraphy Sage, Xixiao River, Shimenkan, Qianguan Alley, and Xinhe Lane, with a total of 62 heritage protection units of varing grades (Fig.2).

① Data source: official website of Shaoxing Municipal People's Government.

Fig.2 Study area.

Full size|PPT slide

The study utilized the ArcGIS Pro platform to construct a spatial analysis grid with cells of 20 m × 20 m each. Spanning from August 1, 2022, to January 31, 2023, the study period supported the comparative analysis between weekdays and weekends/holidays.

4.2 Data Collection and Processing

4.2.1 Weibo Check-in Data

Weibo check-in data include the textual content and timestamp and location information generated when users perform the function of check-ins. The study collected Weibo check-in data within the study area from August 1, 2022, to January 31, 2023. Each piece of data included user ID, posted text, time of posting, and geographic coordinates (Tab.1). After eliminating duplicate, blank or irrelevant data (such as purchasing agents and house rentals) and unrelated web links, 8, 151 pieces of valid check-in data were obtained.

Tab.1 Examples of Weibo check-in data

User	Textual data	Release time	Longitude	Latitude
A	"Xu Wei Art Museum is a surprise in Shaoxing. It has a sense of ideal pureness, and I like the building so much."	2022-08-12	120.574°E	29.998°N
B	"Extreme passion is spontaneous! Everyday on the road of enthusiasm, with light in the eyes and a beautiful future..."	2022-10-01	120.582°E	30.004°N
C	"Walking around the Cangqiao Zhijie Street, where I lived for two years many, many years ago. This is an old street, every morning the seniors chatting on the street and used to be busy all day long. Now there are various small shops including several cafes, with quite a charming style."	2022-11-17	120.573°E	30.005°N
D	"Kong Yiji has not been the figure of sadness, but has become a celebrity for fennel beans."	2023-01-07	120.578°E	29.995°N

4.2.2 Data of HUL Characteristics

The data of HUL characteristics included POI data, historical heritage data, and remote sensing imagery data. POI data, collected from Amap from October 2 to 4, 2022, covered recreational and entertainment facilities, medical service facilities, and transportation facilities. Historical heritage data included information of the protection grade, construction era, and geographic location of the heritage sites within the study area, sourced from the Shaoxing Historical and Cultural City Protection Plan (2021–2035). The study also obtained remote sensing data with a resolution of 0.5 m from Google Earth, taken on June 12, 2021, to identify land cover types and spatial layouts within the study area. With the aid of ArcGIS Pro platform, six land cover types were identified: green space, water body, bare land, building, road, and impervious surface, and 20 ~ 30 grid cells of each type were randomly selected for validation.

4.3 Research Methods

4.3.1 Measurement of Public Sentiments

The study employed Python to perform sentiment analysis of the textual content of the Weibo check-in data. First, each piece of text was segmented using the Jieba library, Python's Chinese word segmentation module^[40], and the frequency of each word was counted. Then, the study utilized the open-source BosonNLP sentiment dictionary^[41] to determine the value assignment of the sentiments. This dictionary contains approximately 120, 000 sentiments, and each of them with an associated score (Tab.2). Positive or negative scores reflect the nature of the sentiment, i.e., positive, neutral or negative sentiments, and the absolute value of the score indicates the intensity of the sentiment. Using the geographic location information from the check-in data, the study applied the Inverse Distance Weighting (IDW) interpolation^[42] method to perform spatial interpolation of the sentiment scores for each piece of check-in data, through which the distribution of public sentiments within the study area can be obtained. The calculation formula is as follows:

Tab.2 Examples of word value assignment by the BosonNLP sentiment dictionary

Category	Word	Assigned value
Positive	Content	2.1
	Joyful	2.6
	Happy	2.6
	Peaceful	0.8
Negative	Irritable	–4.4
	Disturbed	–6.5
	Scared	–4.1
	Unscrupulous	–4.1
Neutral	According to	0.0
	Phenomenon	0.0
	South Zone	0.0
Adverb of degree	Remarkably	1.8
	Very	1.8
	A little	0.7
Deactivated word	Too; also; thus; besides	Eliminated, no assigned value

NOTEThe research adopted the already assigned values by the BosonNLP sentiment dictionary.

(1)$\hat{Z}\left(X_j\right)=\sum _{i=1}^N{\alpha }_{i}Z\left(x_i\right),$

where $\hat{Z}\left(X_j\right)$ is the predicted score of the sentiment at location j; Z(x_i) is the measured score of the sentiment at location i; and α_i is the reciprocal of the distance between the predicted location j and the known location i.

4.3.2 Measurement of the Spatio-temporal Distribution of Public Sentiments

The visualization method of Space-Time Cube was employed to reveal the spatio-temporal differentiation patterns of public sentiments within the study area. Space-Time Cube is a three-dimensional data cube, and each cube unit has a fixed position (x, y, t), where t represents the time step, i.e., the difference between two consecutive time points and (x, y) represents the spatial location of the unit^[43][44] (Fig.3). Referring to the grid cells, the cube unit of the Space-Time Cube set to a length of 20 m and the time step set to one day.

Fig.3 Space-Time Cube.

Full size|PPT slide

Subsequently, the study used the Getis-Ord Gi* index for hotspot clustering analysis. This index can reflect the differentiation of the levels of public sentiments between a given zone and its surrounding areas, so that to identify hot and cold spots, i.e., the clustered and dispersed zones of public sentiments. The calculation formula is as follows^[43]:

(2)$G_i^{\ast }=\frac{\sum _{j=1}^n{w}_{i,j}{x}_j-\overline{X}\sum _{j=1}^n{w}_{i,j}}{S\sqrt{\frac{\left[n\sum _{j=1}^n{w}_{i,j}^2-{\left(\sum _{j=1}^n{w}_{i,j}\right)}^2\right]}{n-1}}},$

where i represents a given grid cell; j represents all its surrounding grid cells; x_j is the average score of public sentiments within j; w_{i, j} is the spatial weight between grid cells i and j, calculated using the default distance-based algorithm on the ArcGIS Pro platform to measure the spatial correlations of public sentiments; and n is the total number of grid cells. When the Getis-Ord Gi* result is positive and significant, the higher the value is, the closer the positive sentiments cluster; conversely, when the result is negative and significant, the lower the value is, the closer the negative emotions cluster.

4.3.3 Analysis of the Influencing Mechanism of HUL Characteristics on Public Sentiments

The study used the GWR model to examine the influencing mechanism of HUL characteristics on public sentiments. Unlike traditional global regression models (e.g., Ordinary Least Squares), GWR model can effectively explain local differences in regression coefficients^[45], i.e., supporting the refined analysis of the relationships between the HUL characteristics and public sentiments at a smaller scale and to identify the ones significantly affecting public sentiments. The model expression is:

(3)$y_i=\sum _{j=1}^k{\beta }_k\left(u_i,v_i\right)x_{ij}+{\epsilon }_i,$

where y_i is the average score of public sentiments at the unit i within the Space-Time Cube; (u_i, v_i) is the geographical coordinates of unit i; x_ij is the independent variable j for public sentiment score of unit i, i.e., HUL characteristics in this study (see Section 6); β_k (u_i, v_i) is the regression coefficient for the k-th independent variable of unit i; and ε_i is the residual error.

5 Spatio-temporal Differentiation Patterns of Public Sentiments in the Shaoxing Ancient City

Analysis results (Tab.3) indicate that public sentiments in the Shaoxing ancient city had an average score of 6.983, and a mean intensity value of 7.160. This displayed an overall positive sentiments among the public, and the negative sentiments did not dominate, suggesting that the public in the study area tended towards positive emotional experiences.

Tab.3 Descriptive statistics of public sentiments

Variable	Mean	SD	Minimum	Maximum
Score of public sentiments	6.983	6.664	–22.600	65.896
Intensity of public sentiments	7.160	6.474	0.000	65.896

The study further conducted a detailed analysis of the number and temporal variation of the Weibo posts, and found that there were significant peaks in early October 2022 and late January 2023 (during the National Day holiday and the eve of the Spring Festival, respectively). While, there was a noticeable decrease around the end of December 2022, possibly due to the pandemic outbreak in China at that time, when the public's outings and social media usage were reduced. The fluctuations in posting volume were relatively small in other months. Additionally, the results show that the number of posts on weekends/holidays was significantly higher than that on weekdays (Fig.4).

Fig.4 Analysis of the number of Weibo posts per day.

Full size|PPT slide

The changes of the daily average score of public sentiments are shown as Fig.5. It found that although on certain dates (such as around September 1) the intensity of negative sentiments surpassed that of positive sentiments, the daily average score of public sentiment remained positive. This could be attributed to the fact that, despite the higher intensity of negative sentiments on those days, the number of positive sentiment posts far exceeded that of negative sentiment posts. This numerical disparity resulted in positive daily average scores even on days with stronger negative sentiments.

Fig.5 Analysis of the daily average score of public sentiments.

Full size|PPT slide

The study also visualized the daily average scores of public sentiments. The results indicate significant clusters of public sentiments and a notable spatio-temporal heterogeneity between weekdays and weekends/holidays. Specifically, on weekdays, zones with stronger positive sentiments were mainly distributed around the entrance of ancient city on the Shangda Road in the northeastern part of the study area, the northern part of the Yuezi Town neighborhood, and along Jiefang Road; strong negative sentiments were found around the Wanghua Residential Community in the southern part of the study area, around the Jianhu-Xincun Residential Community in the southwestern part of the study area, and around the west side of the Former Residence of Calligraphy Sage neighborhood (Fig.6). On weekends/holidays, stronger positive sentiments were primarily found in the northeastern corner of Yuezi Town neighborhood, Bazi Bridge neighborhood, and the western part of Lu Xun's Former Residence neighborhood; meanwhile, the negative sentiments were scattered around the Wanghua Residential Community, the west side of the Former Residence of Calligraphy Sage neighborhood, and the Huayuan-Xincun Residential Community in the west (Fig.7).

Fig.6 Spatial distribution of public sentiments on weekdays.

Full size|PPT slide

Fig.7 Spatial distribution of public sentiments on weekends/holidays.

Full size|PPT slide

A hot and cold spot analysis of the Space-Time Cubes of daily public sentiments yielded 54, 448 cube units, which were visualized using the ArcGIS Pro platform, with snapshots taken on the 10th, 20th, and 30th of each month (Fig.8). The results showed that during August and September, 2022, stronger positive sentiment clusters first appeared in the Qianguan Alley neighborhood. Over time, the neighborhoods of Bazi Bridge and the Former Residence of Calligraphy Sage also showed significant clusters of stronger positive sentiments. As the National Day holiday approached, on September 30, the Lu Xun's Former Residence neighborhood and featured heritage sites like Shen Garden in the southern part of the study area exhibited more clustering of stronger positive sentiments. From October 1 to 10, other locations such as the Shimenkan neighborhood and the Jindi Intime City in the southwestern part of the study area also showed an increase of the hot spots of positive sentiments, aligning with the joy brought by holiday tourism. During November and December, the clustering of positive sentiments declined, concentrating in areas such as the neighborhoods of Shimenkan, the Former Residence of Calligraphy Sage, and Lu Xun's Former Residence, and the entrance of ancient city on the Shangda Road. In early January 2023, although there was a noticeable rise of the clustering of positive sentiments, particularly around the Lu Xun's Former Residence neighborhood, public sentiments were negative overall.

Fig.8 Hot and cold spot analysis of public sentiments.

Full size|PPT slide

6 Influencing Mechanism of HUL Characteristics on Public Sentiments

6.1 Selection of Influencing Factors

This study selected a total of 11 variables of HUL characteristics as explanatory variables from the dimensions of heritage intrinsic value, urban functional value, and urban landscape value, to explore its influencing mechanisms on public sentiments (Tab.4).

Tab.4 Explanatory variables for the impact of public sentiments

	Variable	Description	Source
Dependent variable
	Score of public sentiments (weekday)	Score of public sentiments on weekdays	—
	Score of public sentiments (weekend/holiday)	Score of public sentiments on weekends/holidays	—
Independent variable
Heritage intrinsic value	Heritage level	Assigned value of the level of heritage conservation units	Refs. [32] [33]
	Heritage age	Assigned value of the age of heritage conservation units	Refs. [32] [33]
	Natural asset value	DEM calculation of slope	Ref. [46]
Urban functional value	Building height	Overall height of the building (m)	Ref. [48]
	Internet popularity	Count of internet posts	Refs. [26] [37]
	Density of recreational and entertainment facility	Distribution density of recreational and entertainment POI (per m2)	Refs. [26] [37] [47]
	Density of medical service facility	Distribution density of medical service POI (per m2)	Refs. [26] [37] [47]
	Density of transportation facility	Distribution density of transportation POI (per m2)	Refs. [26] [37] [47]
Urban landscape value	Openness of water body	Area of water body (m2)	Ref. [27]
	Openness of green space	Area of green space (m2)	Ref. [27]
	Degree of mixed land use	∑Piln(Pi)Pi is the area ratio of the land use type i to the grid	Refs. [27] [47]

1) Heritage intrinsic value: in this study, heritage level, heritage age, and natural asset value were selected as the evaluation indicators. Heritages within the study area were assigned values according to their protection levels—national (5), provincial (4), municipal (3), district (2), and ordinary buildings within the ancient city (1). Heritage age refers to the construction eras of the historical structures, with assigned values from 5 to 1 if the heritage was initially constructed earlier than the Tang Dynasty (before the year of 618), in Tang and Song Dynasties (618–1279), Yuan and Ming Dynasties (1271–1644), Qing Dynasty (1644–1911), and modern times, respectively. Natural asset value, the geographical and environmental foundation influencing the formation of the ancient city, was represented with the DEM (Digital Elevation Model) of the terrain, referring to the research of Yuqi Lu et al.^[46]: the variation of terrain relief defines the appearance of natural landscapes, where greater slopes are more likely to shape magnificent landscapes and smaller slopes may present gentle and pleasant sceneries.

2) Urban functional value: drawing from existing studies that use density, mix degree, and urban form to represent the built-up environment of a city^[26][47][48], this study evaluated the urban functional value of HUL with indicators of building height, internet popularity, density of recreational and entertainment facility, density of medical service facility, and density of transportation facility.

3) Urban landscape value: considering the interwoven waterways in the Shaoxing ancient city, the study used indicators such as openness of water body, openness of green space^[27], and degree of mixed land use^[49] to measure the urban landscape value of HUL. This research adopted the measurement methods of urban landscape features by Lingqiang Kong et al.^[27].

6.2 Analysis of Model Reliability

This study employed ArcGIS Pro analytical tools to perform the model credibility analysis on the scores of public sentiments on weekdays and weekends/holidays with the aforementioned 11 explanatory variables. Variables with a variance inflation factor (VIF) greater than 7.5 were excluded to eliminate the effects of multicollinearity. The maximum VIF value among the variables on weekdays was 1.42, and that on holidays was 1.30, indicating that the selected variables were reasonably selected.

The study performed a spatial autocorrelation test, yielding a global Moran's I index of 0.718 for the scores of public sentiments on weekdays, with Z-score of 113.469 (P < 0.05), passing the significance test. For weekends/holidays, the global Moran's I index was 0.812, with the Z-score of 171.508 (P < 0.05), also passing the significance test. These results indicate a spatially significant positive self-correlation of the scores of public sentiments.

Based on the above steps, the study constructed GWR models for weekdays and weekends/holidays and the preliminary results (Tab.5) suggested a good fit of the models, demonstrating that both of them can effectively explain the variations in public sentiments.

Tab.5 Results of GWR models

	R2	RADJ2	AIC	σ2	SD	Pseudo-t statistics correction key values
Weekday	0.8753	0.8677	7163.8253	1.3706	1.2922	3.4140
Weekend/holiday	0.9014	0.8954	6836.8019	1.1835	1.1159	3.4134

6.3 Analysis of GWR Results

The regression coefficients for each variable in the GWR models are shown in Tab.6 and Tab.7. If both the mean and median values are positive, it indicates a positive effect of the explanatory variable on public sentiments; if both are negative, it indicates a negative effect.

Tab.6 Results of GWR model coefficients for public sentiments on weekdays

Explanatory variable		Mean	SD	Minimum	Median	Maximum
Heritage intrinsic value	Heritage level	0.7113	0.4162	–12.2628	0.6055	20.2484
	Heritage age	1.1873	0.5980	–14.9569	0.5463	34.6736
	Natural aesthetic value	0.0269	0.0377	–0.7311	0.0121	1.3360
Urban functional value	Building height	0.0010	0.0032	–0.0380	0.0001	0.0508
	Internet popularity (weekday)	0.0007	0.0005	–0.0458	0.0001	0.1445
	Density of recreational and entertainment facility	–0.0002	0.0048	–0.3938	0.0004	0.1557
	Density of medical service facility	–0.0876	0.0268	–1.4089	–0.0480	1.0971
	Density of transportation facility	0.0430	0.0169	–0.7531	0.0218	0.6314
Urban landscape value	Openness of water body	0.0005	0.0010	–0.0109	0.0004	0.0131
	Openness of green space	0.0001	0.0006	–0.0088	0.0001	0.0124
	Degree of mixed land use	0.0154	0.1506	–1.7635	0.0396	1.6686

Tab.7 Results of GWR model coefficients for public sentiments on weekends/holidays

Explanatory variable		Mean	SD	Minimum	Median	Maximum
Heritage intrinsic value	Heritage level	0.0542	0.3853	–30.9836	0.0619	12.8814
	Heritage age	0.5094	0.5495	–27.6664	0.1342	48.1832
	Natural aesthetic value	–0.0059	0.0351	–0.6519	–0.0105	1.0412
Urban functional value	Building height	–0.0018	0.0029	–0.0692	–0.0008	0.1014
	Internet popularity (weekend/holiday)	–0.0016	0.0006	–0.0887	–0.0002	0.0772
	Density of recreational and entertainment facility	–0.0009	0.0044	–0.7486	0.0007	0.2605
	Density of medical service facility	–0.0147	0.0248	–1.8810	0.0048	1.5385
	Density of transportation facility	0.0081	0.0156	–0.7623	–0.0004	0.9051
Urban landscape value	Openness of water body	0.0004	0.0000	–0.0046	0.0001	0.0049
	Openness of green space	0.0005	0.0006	–0.0086	0.0001	0.0063
	Degree of mixed land use	–0.0034	0.1400	–1.6072	0.0068	1.0713

For the intrinsic value, heritage level and heritage age both showed a significant positive correlation with public sentiments, indicating that heritages of higher protection levels and longer history would have a positive effect on public sentiments. However, natural asset value showed a negative effect on weekdays, possibly because on weekdays the public may prioritize work-related environmental factors, such as traffic convenience and proximity to workplaces, and they might put less attention to appreciating natural sceneries under daily work pressure.

In terms of urban functional value, compared with weekdays, recreational and entertainment facilities and medical service facilities had a weaker effect on weekends/holidays. This might be because the demand for urban functional features is not as much as weekdays. The negative effect of building height and internet popularity on public sentiments was also weaker on weekends/holidays, possibly indicating reduced attention to these HUL characteristics.

For the urban landscape value, the effect of the openness of water body and that of green space on public sentiments was relatively low on both weekdays and weekends/holidays, but this does not imply that these factors are not beneficial to the public's psychological health; instead, this may point to a more subtle influencing mechanism, where the aesthetic and recreational values of blue and green spaces may have a long-term, imperceptible positive effect on public sentiments. The degree of mixed land use showed a positive effect on weekdays but a negative effect on weekends/holidays, suggesting that people might prefer simpler and more direct relaxing experiences during weekends/holidays, while areas with higher hybrid land use might lead to negative sentiments.

To further investigate the spatial heterogeneity of the explanatory variables' effects, the study ranked the explanatory variables based on their absolute value of the medians of regression coefficients, for both weekday and weekend/holiday GWR models. Five explanatory variables with relatively significant benefits on public sentiments on both models—heritage level, heritage age, natural asset value, density of medical service facility, and the degree of mixed land use—were identified, and the regression coefficient of each grid cell was visualized (Fig.9).

Fig.9 Spatio-temporal differentiation of the impact of five HUL characteristics on public sentiments.

Full size|PPT slide

1) Heritage level: 61.10% and 52.14% of the grid cells in the weekday and weekend/holiday GWR models, respectively, displayed positive regression coefficients, indicating a positive correlation between heritage level and public sentiments in these cells (i.e. leading to positive sentiments). On weekdays, this positive effect was mainly found around the neighborhoods of Yuezi Town, the Former Residence of Calligraphy Sage, and the Bazi Bridge, as well as the Shen Garden, which are the daily residential and recreational zones for the citizens. On weekends/holidays, it saw a significant increase in the number of grid cells where heritage sites were located showing a positive effect (Fig.9); however, there was a noticeable difference in the regression coefficients of heritage level around the Lu Xun's Former Residence neighborhood between weekdays and weekends/holidays. The Lu Xun's Former Residence neighborhood, as a famous tourism destination that attracts the large number of visitors during weekends and holidays, showed a significant positive correlation between heritage level and public sentiments, which locally reached up to 1.76. While, on weekdays, probably due to commercial activities or management regulation measures, the neighborhood showed a significant negative correlation between heritage level and public sentiments, which locally reached up to –12.26. This suggests that the HUL in this study area still needs to strike a balance between tourists' tourism demands and people's needs of daily commercial activities by, for example, taking flexible management measures on different days.

2) Heritage age: 60.85% and 52.15% of the grid cells in the weekday and weekend/holiday GWR models, respectively, exhibited positive regression coefficients, indicating a positive correlation between heritage age and public sentiments in these cells. This influencing pattern is similar to that of heritage level. On both weekdays and weekends/holidays, positive effects were concentrated in the Bazi Bridge neighborhood and several blocks in the southern part of the study area, where various cultural heritage sites spanning multiple historical periods sit. However, around the neighborhoods of the Former Residence of Calligraphy Sage, Yuezi Town, and Xixiao River, more grid cells with negative effects were found (Fig.9). This might be because of the poor preservation status of those historical neighborhoods and the shortage of maintenance and supporting facilities, which would affect the visitors' experiences.

3) Natural asset value: 58.56% and 41.99% of the grid cells in the weekday and weekend/holiday GWR models, respectively, showed positive regression coefficients. On weekdays, typical Jiangnan water town landscape areas (such as Bazi Bridge neighborhood) and mountainous areas (such as the Yuezi Town neighborhood and Tashan Mountain scenic area) exhibited significant positive effects on public sentiments. However, during weekends and holidays, water town landscape areas still showed a significant positive effect, but the zones of mountainous landscapes showed a negative effect on public sentiments, possibly due to increased tourist traffic and crowded visiting experience (Fig.9).

4) Density of medical service facility: 37.45% and 51.45% of the grid cells in the weekday and weekend/holiday GWR models, respectively, exhibited positive regression coefficients. Overall, the density of medical service facility showed a significant negative effect with public sentiments. However, in central areas of the ancient city, such as Zhongxing Middle Road and Guojin Joy City, where residential areas dominated, the density of medical service facility had a positive effect on public sentiments. This suggests that in densely populated residential areas, a reasonable distribution of medical service facilities can enhance positive public sentiments (Fig.9).

5) Degree of mixed land use: 58.90% and 51.69% of the grid cells in the weekday and weekend/holiday GWR models, respectively, displayed positive regression coefficients. Areas such as the neighborhoods of the Lu Xun's Former Residence and Shimenkan, and Shen Garden showed positive effects on both weekdays and weekends/holidays. The neighborhoods of the Former Residence of Calligraphy Sage and Xixiao River also showed positive effects during weekends/holidays, demonstrating that for relaxing times multi-functional historical and cultural sites can effectively enhance the public's positive sentiments (Fig.9).

7 Conclusions and Discussion

As an embodiment of a city's cultural heritages and historical memories, HUL can evoke collective emotional resonance and a sense of identity, thereby enhancing their happiness and satisfaction. Based on HUL and Emotional Geography theories, this study proposed the HCS analysis framework—consisting of three dimensions of heritage intrinsic value, urban functional value, and urban landscape value—and explored the spatio-temporal patterns of public sentiments and the influencing mechanisms of HUL characteristics on public sentiments in the Shaoxing ancient city. The results show that different HUL characteristics had played varied influencing mechanisms on public sentiments, and the effects of same HUL characteristics on public sentiments also vary between weekdays and weekends/holidays and among different HULs. On weekends/holidays, public sentiments were more influenced by the intrinsic value factors of HUL (e.g., heritage level, heritage age), whereas on weekdays, they were more affected by urban functional value factors (e.g., density of transportation facilities), and urban landscape value factors (e.g., degree of mixed land use) played a greater role in arousing people's positive sentiments.

It is important to note that the influence of HUL characteristics on public sentiments is not static. Urban designers should propose more targeted development planning and policies based on the local impact characteristics of HUL on public sentiments. This aligns with the core idea of HUL, which is to balance HUL preservation with urban development through comprehensive urban planning and management measures, thereby achieving a sustainable urban environment and enhancing people's well-being.

Despite the preliminary exploration of the relationship between HUL characteristics and public sentiments, this study has some limitations. First, existing research indicates that users who actively post information on social media platforms may be more inclined to make extreme or intense emotional expression^[50]. This may result in a disparity between the expressed sentiments from the Weibo data with people's visiting experiences in real world. Future research need to combine with multiple data sources and methods, such as questionnaires and mobile application data, to obtain more comprehensive and more accurate data about the public's sentiments. Second, while this study examined HUL characteristics from the dimension of heritage intrinsic value, urban functional value, and urban landscape value, it is still challenging to fully capture the complexity and diversity of HUL. Future research could consider more potential influencing factors, such as socio-economic and individual factors, combined with relevant theories and technological methods, to better understand the influencing mechanisms of HUL characteristics on public sentiments. Additionally, the findings of this study are targeted to the Shaoxing ancient city. Future research is expected to conduct comparative studies with more cases in other regions, further verifying the universality and reliability of the findings. Simultaneously, continuous studies could also be conducted to track the long-term influencing mechanisms and evolutionary patterns of HUL characteristics on public sentiments in the Shaoxing ancient city.

References

[1] UNESCO. (2019). The UNESCO recommendation on the Historic Urban Landscape.

[2] Li, H. , & Yang, L. (2019) Research on the management tools of Historic Urban Landscape—Historic townscape characterisation. Chinese Landscape Architecture, 35 (35), 54– 58.

[3] & Lazarus, R. S. (1982) Thoughts on the relations between emotion and cognition. American Psychologist, 37 (39), 1019– 1024.

[4] Li, Y. , Chen, X. , Liu, P. , & Huang, G. (2021) Research on image perception of heritage tourist destinations from the three-dimensional perspective of "cognition–emotion–integrity": Taking Xiangjiang ancient town group as an example. Human Geography, 36 (35), 167– 176.

[5] & Taylor, K. (2023) Historic Urban Landscape paradigm—A tool for balancing values and changes in the urban conservation process. Landscape Architecture Frontiers, 11 (33), 96– 104.

[6] Zhang, S. , & Zhen, X. (2017) From historical features preservation to urban landscape management—Based on the Historic Urban Landscape approach. Landscape Architecture, (36), 14– 21.

[7] Zhou, X. , Sun, Z. , & Liu, X. (2022) Research on the spatial perception and recreational experience of Historic Urban Landscape supported by the technology of digital footprint. New Architecture, (31), 10– 15.

[8] Li, H. , Li, M. , Zou, H. , Zhang, Y. , & Cao, J. (2023) Urban sensory map: How do tourists "sense" a destination spatially?. Tourism Management, (397), 104723.

[9] Zhou, X. , Liu, Z. , & Shi, J. (2022) Evaluation on visual landscape perception under the support of digital footprint: A case study of Historic Urban Landscape of the Qinhuai River area. Landscape Architecture, 29 (39), 18– 25.

[10] Ginzarly, M. , Roders, A. P. , & Teller, J. (2019) Mapping historic urban landscape values through social media. Journal of Cultural Heritage, (336), 1– 11.

[11] Dai, J. , Huang, X. , Sun, J. , An, H. , & Guo, Q. (2023) Research on the formation mechanism of tourist cultural identity in National Cultural Park—Taking the iconic natural landscapes of the Yellow River National Cultural Park as an example. Tourism Tribune, 38 (31), 31– 41.

[12] Taylor, K. (2008, September 29–October 4). Landscape and memory: Cultural landscapes, intangible values and some thoughts on Asia[Conference or Workshop Item]. 16th ICOMOS General Assembly and International Symposium: 'Finding the Spirit of Place—between the Tangible and the Intangible'. Quebec, Canada.

[13] Liu, T. , Butler, R. J. , & Zhang, C. (2019) Evaluation of public perceptions of authenticity of urban heritage under the conservation paradigm of Historic Urban Landscape—A case study of the Five Avenues Historic District in Tianjin, China. Journal of Architectural Conservation, 25 (33), 228– 251.

[14] & Hoteit, A. (2015) Role of the landscape in the preservation of collective memory and the enhancement of national belonging. Canadian Social Science, 11 (33), 1– 8.

[15] Bonadei, R. , Cisani, M. , & Viani, E. (2017) City walls as Historic Urban Landscape: A case study on participatory education. Journal of Tourism, Culture and Territorial Development, 8 (37), 75– 88.

[16] Anderson, K. , & Smith, S. (2001) Editorial: Emotional geographies. Transactions of the Institute of British Geographers, 26 (31), 7– 10.

[17] Zhu, H. , & Gao, Q. (2015) Review on "emotional turn" and emotional geographies in recent western geography. Geographical Research, 34 (37), 1394– 1406.

[18] Yang, L. , Duarte, C. M. , & Ciriquián, P. M. (2022) Quantifying the relationship between public sentiment and urban environment in Barcelona. Cities, (3130), 103977.

[19] Lazarus, R. S. (1991). Emotion and Adaptation. Oxford University Press.

[20] Zhang, X. (2019). Emotion recognition via passive sensing [Doctoral dissertation]. Nanjing University.

[21] Li, Y. , Gao, X. , Yang, L. , & Guo, J. (2020) Research on walking environment satisfaction of residents based on ordered probit model—A case study of Gulangyu. Chinese Landscape Architecture, 36 (311), 90– 94.

[22] Kerr, M. L. , Rasmussen, H. F. , Smiley, P. A. , Buttitta, K. V. , & Borelli, J. L. (2021) The development of toddlers' emotion regulation within the family system: Associations with observed parent-child synchrony and interparental relationship satisfaction. Early Childhood Research Quarterly, (357), 215– 227.

[23] Benita, F. , Bansal, G. , & Tunçer, B. (2019) Public spaces and happiness: Evidence from a large-scale field experiment. Health & Place, (356), 9– 18.

[24] Rodríguez-Ibánez, M. , Casánez-Ventura, A. , Castejón-Mateos, F. , & Cuenca-Jiménez, P.-M. (2023) A review on sentiment analysis from social media platforms. Expert Systems with Applications, (3223), 119862.

[25] Mitchell, L. , Frank, M. R. , Harris, K. D. , Dodds, P. S. , & Danforth, C. M. (2013) The geography of happiness: Connecting Twitter sentiment and expression, demographics, and objective characteristics of place. PLoS One, 8 (35), e64417.

[26] Shan, Z. , An, Y. , Yuan, M. , & Huang, Y. (2022) Study on spatial and temporal distribution characteristics and influencing factors of urban residents' sentiment based on Weibo data: A case study of the main urban area of Wuhan. Urban Development Studies, 29 (38), 69– 76.

[27] Kong, L. , Liu, Z. , Pan, X. , Wang, Y. , Guo, X. , & Wu, J. (2022) How do different types and landscape attributes of urban parks affect visitors' positive emotions?. Landscape and Urban Planning, (3226), 104482.

[28] Wang, Y. , Chew, A. W. Z. , & Zhang, L. (2022) Deep learning modeling of public's sentiments towards temporal evolution of COVID-19 transmission. Applied Soft Computing, (3131), 109728.

[29] & Amarasinghe, A. (2023) Public sentiment in times of terror. Journal of Development Economics, (3162), 103058.

[30] Ginzarly, M. , Houbart, C. , & Teller, J. (2019) The Historic Urban Landscape approach to urban management: A systematic review. International Journal of Heritage Studies, 25 (310), 999– 1019.

[31] Jian, J. , Zhen, F. , Xi, G. , & Zhai, Q. (2016) A review on emotional geography: Its progress and enlightenment. World Regional Studies, 25 (32), 123– 136.

[32] Lei, Z. , Luo, J. , & Cao, L. (2021) Quantitative research on historic city landscape value assessment based on AHP-GIS—Taking Taiyuan, a historic and cultural city, as an example. Journal of Taiyuan University of Technology, 52 (34), 630– 637.

[33] Shoval, N. , Schvimer, Y. , & Tamir, M. (2018) Tracking technologies and urban analysis: Adding the emotional dimension. Cities, (372), 34– 42.

[34] Oren, G. , Shani, A. , & Poria, Y. (2021) Dialectical emotions in a dark heritage site: A study at the Auschwitz Death Camp. Tourism Management, (382), 104194.

[35] & Jimura, T. (2011) The impact of world heritage site designation on local communities—A case study of Ogimachi, Shirakawa–mura, Japan. Tourism Management, 32 (32), 288– 296.

[36] Su, M. , & Wall, G. (2010) Place attachment and heritage tourism at the Great Wall. Journal of China Tourism Research, 6 (34), 396– 409.

[37] Medina-Viruel, M. J. , López-Guzmán, T. , Gálvez, J. C. P. , & Jara-Alba, C. (2019) Emotional perception and tourist satisfaction in world heritage cities: The Renaissance monumental site of Úbeda and Baeza, Spain. Journal of Outdoor Recreation and Tourism, (327), 100226.

[38] Liang, J. , Jiang, S. , & Tao, L. (2020) Empirical research on the relationship between spatial-temporal behavior and emotion of tourists based on semantic analysis of online travel blogs and GIS visualization: A case of Beijing. Human Geography, 35 (32), 152– 160.

[39] Shaoxing Wuhan Municipal Bureau of Natural Resources and Planning. (2023, October 23). Public notice of draft of Shaoxing historical and cultural city protection plan (2021–2035).

[40] Zhu, Y. , & Jing, J. (2019) Chinese word segmentation technology based on Python language. Communications Technology, 52 (37), 1612– 1619.

[41] Wang, Y. , Chi, Y. , Xu, J.-H. , & Yuan, Y. (2022) Consumers' attitudes and their effects on electric vehicle sales and charging infrastructure construction: An empirical study in China. Energy Policy, (3165), 112983.

[42] Wan, P. , Qiu, F. , Xiang, F. , Wang, J. , Pan, C. , Zhu, X. , & Xu, L. (2022) Study on the Spatio-temporal Variation and Prediction Model of PM2.5 in Kunming. Environmental Science Survey, 41 (32), 47– 51.

[43] Wang, H. , Chen, R. , Wei, S. , Gui, L. , & Fei, T. (2020) Mining emotional geography features based on Chinese Weibo data. Geomatics and Information Science of Wuhan University, 45 (35), 699– 708.

[44] Zhang, Q. , Xu, D. , & Ding, Y. (2021) Spatio-temporal pattern mining of the last 40 years of drought in China based on SPEI index and spatio-temporal cube. Agricultural Research in the Arid Areas, 39 (33), 194– 201.

[45] Xue, M. , Wang, C. , Zhao, J. , & Li, M. (2020) Spatial differentiation pattern and influencing factors of tourism economy in the Yellow River Basin. Economic Geography, 40 (34), 19– 27.

[46] Tao, J. , Dong, P. , Zhou, N. , Lu, Y. , Gong, X. , & Ma, X. (2022) Study on spatial distribution characteristics and influencing factors of historical and cultural towns and villages in the Yangtze River Economic Belt. Geography and Geo-Information Science, 38 (33), 66– 73.

[47] Wang, C. , Wang, B. , Wang, Q. , & Lei, Y. (2023) Nonlinear associations between urban vitality and built environment factors and threshold effects: A case study of central Guangzhou City. Progress in Geography, 42 (31), 79– 88.

[48] Ta, N. , Zeng, Y. , Zhu, Q. , & Wu, J. (2020) Relationship between built environment and urban vitality in Shanghai downtown area based on big data. Scientia Geographica Sinica, 40 (31), 60– 68.

[49] Zhu, J. , Fan, Y. , & Fan, F. (2018) Factors affecting commute happiness of residents in big cities: A case study of Xi'an, China. Urban & Rural Planning, (33), 43– 53.

[50] Fan, R. , Zhao, J. , Chen, Y. , & Xu, K. (2014) Anger is more influential than joy: Sentiment correlation in Weibo. PLoS ONE, 9 (310), e110184.

RIGHTS & PERMISSIONS

© Higher Education Press 2024