Spatio-Temporal Patterns and Sentiment Analysis of Ting, Tai, Lou, and Ge Ancient Chinese Architecture Buildings

Abstract

Ting, Tai, Lou, and Ge are types of ancient buildings that represent traditional Chinese architecture and culture. They are primarily constructed using mortise and tenon joints, complemented by brick and stone foundations, showcasing traditional architectural craftsmanship. However, research aimed at conserving, inheriting, and rejuvenating these buildings is limited, despite their status as Provincial Cultural Relic Protection Units of China. Therefore, the aim of this study was to reveal the spatial distribution of Ting, Tai, Lou, and Ge buildings across China, as well as the factors driving differences in their spatial distribution. Tourist experiences and building popularity were also explored. The spatial analysis method (e.g., Standard deviation ellipse and Geographic detector), Word cloud generation, and sentiment analysis, which uses Natural Language Processing techniques to identify subjective emotions in text, were applied to investigated the research issues. The key findings of this study are as follows. The ratio of Ting, Tai, Lou, and Ge buildings in Southeast China to that in Northwest China divided by the “Heihe–Tengchong” Line, an important demographic boundary in China with the ratio of permanent residents in the two areas remaining stable at 94:6, was 94.6:5.4. Geographic detector analysis revealed that six of the seven natural and socioeconomic factors (topography, waterways, roads, railways, population, and carbon dioxide emissions) had a significant influence on the spatial heterogeneity of these cultural heritage buildings in China, with socioeconomic factors, particularly population, having a greater influence on building spatial distributions. All seven factors (including the normalized difference vegetation index, an indicator used to assess vegetation health and coverage) were significant in Southeast China, whereas all factors were non-significant in Northwest China, which may be explained by the small number of buildings in the latter region. The average rating scores and heat scores for Ting, Tai, Lou, and Ge buildings were 4.35 (out of 5) and 3 (out of 10), respectively, reflecting an imbalance between service quality and popularity. According to the percentages of positive and negative reviews, Lou buildings have much better tourism services than other buildings, indicating a need to improve services to attract more tourists to Ting, Tai, and Ge buildings. Four main types of words were used with high frequency in the tourism reviews collected form Ctrip, a popular online travel platform in China: (1) historical stories; (2) tourism; (3) culture; and (4) cities/provinces. Ting and Tai buildings showed similar word clouds, as did Lou and Ge buildings, with only the former including historical stories. Conversely, landmark was a high-frequency word only in the reviews of Lou and Ge buildings. Specific suggestions were proposed based on the above findings to promote tourism and revive ancient Chinese architecture.

1. Introduction

Cultural heritage encompasses material and spiritual wealth created during the development of human history [1,2] and is typically divided into material/tangible and intangible assets [3,4]. Tangible cultural heritage involves movable (e.g., paintings and manuscripts) and immovable (e.g., ancient sites, buildings, and landscapes) materials, whereas intangible cultural heritage includes folk customs, dance, cultural beliefs, and festivals [4]. Cultural heritage, widely accepted as the most valuable asset in the world, plays a vital role in fostering national identity, social cohesiveness, and cultural confidence, not only by attesting to history, but also by representing the essence of a country [5,6]. Accordingly, cultural heritage is a key area of international research, particularly in relation to conservation, inheritance, and development [7,8,9,10].

As an ancient country with a long history, China boasts an abundant diverse cultural heritage with a wealth of related research [1,11]. Previous research in this field, which has explored material cultural heritage [12,13,14], intangible cultural heritage [15,16,17], and both types combined [1,18,19], has revealed the spatial patterns and driving factors of cultural heritage on national [1,14], provincial [15,20], and urban [12,21] scales. Chinese cultural heritage has also been investigated on more complex regional scales, including Southwest China [18], the land border of China [4], the Jiang–Zhe–Hu region [17], the Beijing–Tianjin–Hebei region [16], natural mountain regions [19], and natural watersheds [13]. The types of cultural heritage analyzed in these studies include, but are not limited to, ancient buildings, canal sites, movable cultural remains, industrial heritage, socialist built heritage, tea and porcelain ancient road culture, traditional sports, entertainment, traditional medicine, and toponymic cultural heritage. However, no previous studies have focused on the ancient Chinese architectural categories of Ting, Tai, Lou, and Ge, which represent a crucial aspect of traditional Chinese architecture and culture [22].

Determining the spatial distribution of cultural heritage and the factors driving these patterns is vital to the conservation and rejuvenation of cultural heritage as well as the integration of culture and tourism [23]. Tourism is an ideal method for revitalizing cultural heritage [24,25] as these assets provide visitors with an experience of local culture, thus promoting cultural exchange and revitalization [26]. Tourism development also boosts economic growth and provides a foundation for protecting cultural heritage [27,28]. Accordingly, cultural tourism has become the dominant form of tourism in China [29]. Visitors predominantly share their travel experiences through social media; such user-generated content then influences potential visitors, who perceive it as more reliable than official information [30,31]. Therefore, exploring user-generated content is crucial for understanding the popularity and service quality of cultural heritage sites.

This study reveals the spatio-temporal distribution and driving factors of Ting, Tai, Lou, and Ge architectural buildings, which belong to the Provincial Cultural Relics Protection Units of China. Moreover, user-generated rating scores (the indicator for the user satisfaction of visiting tourist attractions), heat scores (the indicator for the popularity of tourist attractions), and comments from Ctrip, a popular online travel platform in China [32,33] which allows users to browse, review, and rate tourist attractions, are employed to explore the tourist experience and popularity of Ting, Tai, Lou, and Ge buildings via spatial analysis and sentiment analysis. This study provides experimental evidence for improving tourism strategies at Ting, Tai, Lou, and Ge buildings, and has implications for promoting local economic and cultural revival, as well as the conservation, inheritance, and rejuvenation of ancient Chinese architecture.

2. Data Sources

The data sources of the fundamental information of Ting, Tai, Lou, and Ge buildings (e.g., name and coordinates) and the tourism information (e.g., rating score and reviews) were elaborated on. Moreover, the source of the seven factors driving the spatial patterns of Ting, Tai, Lou, and Ge buildings were shown.

2.1. Ting, Tai, Lou, and Ge Building Information

The characteristics of Ting, Tai, Lou, and Ge architectures are shown in Figure 1. Ting is often an open, wall-free structure that serves as a spot to rest, enjoy shade, or appreciate the scenery; Tai describes open structures that are elevated above the ground; Lou refers to large structures with two or more levels; and Ge is comparable to Lou but has multiple doors and windows on the sides for viewing the surrounding area [22]. They are primarily constructed using mortise and tenon joints, complemented by brick and stone foundations, showcasing traditional architectural craftsmanship.

Buildings with Ting, Tai, Lou, and Ge architectures were filtered from lists of Provincial Cultural Relics Protection Units in China. A total of 482 buildings were analyzed in this study, consisting of 43 Ting, 55 Tai, 277 Lou, and 107 Ge. Information related to the date of construction, address, latitude, and longitude was also collected, along with rating scores, heat scores, and reviews from Ctrip. The latitudes and longitudes were picked from Baidu Maps (https://api.map.baidu.com/lbsapi/getpoint/index.html, accessed on 8 March 2024) or Gaode Maps (https://lbs.amap.com/tools/picker, accessed on 8 March 2024), then converted to the World Geodetic System 1984 (WGS84) using an online tool (https://tool.lu/coordinate/, accessed on 8 March 2024). Rating scores, heat scores, and review data of Ting, Tai, Lou, and Ge buildings were obtained from Ctrip (https://you.ctrip.com, accessed on 8 March 2024) using Octopus Collector software 8.7.2 on 9 August 2024. The rating score reflects the user’s degree of satisfaction with the scenic spot and ranges from 0 to 5. The heat score reflects the popularity of tourist attractions (i.e., Ting, Tai, Lou, and Ge), calculated from the number of user visits, reviews, and so on by Ctrip, and ranges from 0 to 10. For buildings with fewer than 100 reviews, all reviews were collected; for buildings with more than 100 reviews, only 100 reviews were collected. Reviews containing only emoticons and no text were cleaned prior to analysis.

2.2. Analysis of Factors Driving the Spatial Distribution of Buildings

The factors driving the spatial patterns of Ting, Tai, Lou, and Ge buildings were analyzed from natural and socioeconomic perspectives. Natural driving factors included topography (digital elevation model; DEM), the normalized difference vegetation index (NDVI), and the length of waterways. The NDVI is an indicator used to assess vegetation health and coverage. Socioeconomic driving factors included the length of roads, length of railways, population, and carbon dioxide (CO₂) emissions. Greenhouse gas emissions were analyzed from all sectors, including the energy sector, industrial combustion and processes, and transport, to indicate the level of economic development.

DEM data for China were collected from the Resource and Environmental Science Data Platform in Geotiff format with WGS84 projection and a resolution of 250 m (https://www.resdc.cn/data.aspx?DATAID=123, accessed on 15 July 2024). NDVI data, in the China Geodetic Coordinate System 2000 with a spatial resolution of 250 m, were collected from the National Tibetan Plateau Data Center (https://data.tpdc.ac.cn/zh-hans/data/10535b0b-8502-4465-bc53-78bcf24387b3, accessed on 15 July 2024). This dataset is a monthly product that synthesizes the maximum 16-day NDVI value of the MOD13Q1 product from the Aqua/Terra-MODIS satellite sensor [34]. NDVI data were collected from July 2020 to coordinate with population data and because July represents the middle of summer, when both evergreen and deciduous plants are green, thereby avoiding seasonal influences in the dataset. Data were preprocessed by setting the filled value (−3000) to null. Data related to waterways, railways, and roads in China were downloaded in shape file format with a projection of WGS84 from Geofabrik, which offers regional extracts of OpenStreetMap data (https://download.geofabrik.de/asia/china.html, accessed on 15 July 2024). Given the range of road sizes in China, primary, secondary, and tertiary roads were extracted for analysis. Waterways, railways, and roads were projected onto the Krasovsky_1940_Albers coordinate system to calculate road lengths. Chinese population data for 2020 were downloaded from the WorldPop Hub in Geotiff format with WGS84 projection (https://hub.worldpop.org/geodata/summary?id=49919, accessed on 15 July 2024) and a resolution of three arcs (approximately 100 m at the equator). CO₂ emissions for 2020 were downloaded from the Emissions Database for Global Atmospheric Research, provided by the European Commission, in NetCDF (.nc) format with a projection of WGS84 and a resolution of 0.1 × 0.1° (approximately 11.1 × 11.1 km) in units of tons (https://edgar.jrc.ec.europa.eu/gallery?release=v80ghg&substance=CO2&sector=TOTALS, accessed on 15 July 2024).

3. Methods

Figure 2 shows the research flowchart. As could be seen, the standard deviation ellipse, geographic detector, and so on were used to analyze the spatial distribution of Ting, Tai, Lou, and Ge buildings, as well as the factors driving differences in their spatial distribution. In addition, sentiment analysis, kernel density estimation, and word cloud generation which were applied to explore tourist experiences and building popularity were explained.

3.1. Standard Deviation Ellipse

A standard deviation ellipse was applied to evaluate the central, dispersed, and directional trends of the geographic features. In this study, an elliptical azimuth was used to assess the major trend directions of Ting, Tai, Lou, and Ge building distributions. The central coordinates of the ellipse were calculated using Equations (1) and (2), where the long and short axes display the directions of maximum and minimal diffusion, respectively [35].

$$x_{SDE}=\sqrt{{\displaystyle \frac{{\sum }_{i=1}^n{(x_i-\overline{x})}^2}{n}}},$$

(1)

$$y_{SDE}=\sqrt{{\sum }_{i=1}^n{\displaystyle \frac{{(y_i-\overline{y})}^2}{n}}},$$

(2)

Here, $x_{SDE}$ and $y_{SDE}$ are the central coordinates of the ellipse, $x_i$ and $y_i$ are the coordinates of the building, $\overline{\mathrm{x}}$ and $\overline{\mathrm{y}}$ are the average centers of the buildings (i.e., Ting, Tai, Lou, and Ge buildings), and n is the number of buildings.

3.2. Geographic Detector

A geographic detector method (GeoDetector) was developed to measure and attribute spatially stratified heterogeneity, which is a ubiquitous phenomenon in nature whereby data points within a specific stratum are more similar to each other than to data points in other strata [36,37,38]. This measure was used as the dependent variable to quantitatively determine the influence of natural and socioeconomic factors on the spatial distribution of Ting, Tai, Lou, and Ge buildings (Equation (3)).

$$\mathrm{q}=1-{\displaystyle \frac{\sum _{h=1}^L{N}_h{\sigma }_h^2}{{N\sigma }^2}},$$

(3)

Here, q is the q-statistic of the geographical detector, N is the number of buildings; ${\sigma }^2$ is the variance of buildings, $N_h$ and ${\sigma }_h^2$ are the sample size and variance of the h-th type of influencing factors, respectively, and L is the classification/subregion number of buildings. The larger the q-value, the greater the influence of the factor on the spatial distribution of Ting, Tai, Lou, and Ge buildings.

3.3. Kernel Density Estimation

Kernel density estimation is a nonparametric estimation method used to calculate the spatial density of geographic features (i.e., Ting, Tai, Lou, and Ge buildings in this study) and analyze the degree of aggregation and spatial distribution patterns of regional resources [14,16]. The kernel density was estimated using Equation (4):

$$f_n(\mathrm{x})={\displaystyle \frac{1}{nh}}{\sum }_{i=1}^{n}k\left({\displaystyle \frac{{x-x}_i}{h}}\right),$$

(4)

where $f_n$(x) is the estimate of the kernel density, k(${\displaystyle \frac{{x-x}_i}{h}}$) is the kernel function, n is the number of buildings (i.e., Ting, Tai, Lou, and Ge buildings), h is the search radius, and (${x-x}_i$) is the distance from assessment point x to building point x_i.

3.4. Sentiment Analysis and Word Cloud Generation

Sentiment analysis was conducted using SnowNLP, which uses an internal segmentation approach and scoring after segmentation. The scoring method employs the Naïve Bayes technique. To maximize the effectiveness of SnowNLP sentiment analysis, the segmentation process was improved by merging with word segmentation using the Jieba Library, which is a first-class third-party Python package 3.9.2 for effective Chinese-word segmentation. In addition, the Naïve Bayes algorithm employed training data (i.e., Beijing tourist attraction review data on Ctrip) downloaded from China Scientific Data (http://doi.org/10.11922/sciencedb.01623, accessed on 15 July 2024). The training data contained 48,934 comments, including 36,185 positive and 12,749 negative comments. The scoring results for texts range from 0 to 1; the closer the value is to 1, the more positive the sentiment, whereas the closer the value is to 0, the more negative the sentiment. In this study, 0.5 was used as the threshold, representing neutrality, with 0–0.5 indicating negative scores and 0.5–1 indicating positive scores. Finally, word clouds were generated based on the word frequency (using the 200 most frequently occurring words), which were counted after conducting word segmentation on all comments.

4. Results and Discussion

This Section first exhibits the general building characteristics of Ting, Tai, Lou, and Ge buildings, their spatio-temporal patterns, and the driving factors leading to such patterns. Additionally, the distribution of rating and heat scores of Ting, Tai, Lou, and Ge buildings at the provincial level and the sentiment of tourists are shown.

4.1. General Building Characteristics

Table 1. Number of Ting, Tai, Lou, and Ge buildings with and without construction dates and reviews.

	Ting	Tai	Lou	Ge	All
Total	43	55	277	107	482
With construction date	42	54	267	101	464
With reviews	9	15	51	14	89
Percent of buildings with reviews (%)	20.93	27.27	18.41	13.08	18.46
<100 reviews	7	12	40	11	70
≥100 reviews	2	3	11	3	19
Percent of buildings with over 100 reviews (%)	4.65	5.45	3.97	2.80	3.94

shows the number of Ting, Tai, Lou, and Ge buildings with and without the date of construction and reviews on Ctrip. Approximately 18.46% of buildings have reviews on Ctrip, but only 3.94% have more than 100 reviews. The average heat score for the buildings is 3.0. In contrast, the famously popular and scenic Imperial Palace in Beijing has over 100,000 reviews, a rating score of 4.8, and a heat score of 10. This indicates that, in general, Ting, Tai, Lou, and Ge buildings are not sufficiently promoted so do not attract sufficient attention. In the future, the government should intensify publicity efforts for these historical buildings to advance the dissemination of traditional Chinese culture and tourism development.

4.2. Spatio-Temporal Patterns

Figure 3 shows the spatial distribution of Ting, Tai, Lou, and Ge buildings across China, whereas Figure 4 shows the number of buildings in each province. The three provinces of Guangdong, Anhui, and Henan contain the most buildings, with 42, 38, and 36, respectively. Guangdong likely contains the most buildings because traditional culture is most effectively preserved in this province; its ancestral temple culture reflects the respect of the Guangdong people for their ancestors and traditions. Anhui is a cultural province with rich and colorful local customs and unique historical and cultural connotations. Henan is an important birthplace of Chinese civilization, with a rich historical and cultural heritage and important archeological discoveries. In contrast, Shanghai city does not contain any Ting, Tai, Lou, or Ge buildings listed in the Provincial Cultural Relics Protection Units of China. Heilongjiang, Jilin, Ningxia, and Tibet contain one building each. Regarding the different types of buildings, Zhejiang, Guangxi, Guangdong, and Shandong provinces contain the highest number of Ting (11), Tai (16), Lou (35), and Ge (15) buildings, respectively.

Figure 5 shows the standard deviation ellipses of Ting, Tai, Lou, and Ge buildings. The ellipse for Ting buildings is located further southeast than all other ellipses and shows a clear northwest–southeast alignment of the major axis. The ellipses for all buildings and Lou buildings are almost overlaid and vertically oriented. The rotation of the long axis, measured clockwise from the vertex, is 32.53°, 65.15°, 21.99°, 164.40°, and 51.14° for all, Ting, Tai, Lou, and Ge buildings, respectively (

Table 2. Standard deviation ellipse parameters for Ting, Tai, Lou, and Ge buildings.

	Center X	Center Y	XStdDist	YStdDist	Rotation
All buildings	113.44	30.94	7.17	8.04	32.53
Ting	115.36	28.11	7.69	4.35	65.15
Tai	113.02	29.93	5.29	8.61	21.99
Lou	113.39	31.13	7.36	8.02	164.40
Ge	113.01	32.12	9.19	5.84	51.14

). The ellipse for Lou buildings is closest to a circle, indicating an approximately equal distribution of buildings in both south–north and east–west orientations. In contrast, the ellipse for Tai buildings has the largest ratio between the Y-axis and the X-axis, indicating that these buildings are distributed with a south–north alignment. The ellipses for Tai and Lou buildings have longer Y-axes, whereas those for Ting and Ge have longer X-axes.

Figure 6 shows the numbers of Ting, Tai, Lou, and Ge buildings built during different Chinese dynasties. The majority of buildings, 464 out of 482, have information on the date of construction (Table 1). Most buildings were built during the Qing dynasty (275), followed by the Ming dynasty (150). All other dynasties are represented by fewer than 20 buildings. The number of buildings generally decreases with the age of construction, which is likely because newer buildings are better preserved. However, only two buildings were built in modern times; this may reflect the impact of multiple wars on more modern buildings.

4.3. Driving Factors

GeoDetector was used to analyze the factors driving the spatial distribution of Ting, Tai, Lou, and Ge buildings in China. Natural breaks (i.e., Jenks, five clusters) were applied to the data for China, Southeast China, and Northwest China. Southeast and Northwest China were divided by the “Heihe–Tengchong” Line, which is an important demographic boundary in China, with the ratio of permanent residents in the two areas remaining stable at 94:6 for approximately the last 70 years, albeit with slight fluctuations [39]. Specifically, the northwest part includes the Inner Mongolia Autonomous Region (excluding Chifeng City, Tongliao City, and Xing’an League), Gansu Province, the Ningxia Hui Autonomous Region, the Xinjiang Uygur Autonomous Region, Qinghai Province, the Xizang Autonomous Region, the Great Khingan Mountains region of Heilongjiang Province, the Aba Tibetan and Qiang Autonomous Prefecture and Ganzi Tibetan Autonomous Prefecture of Sichuan Province, the Nujiang Lisu Autonomous Prefecture and Diqing Tibetan Autonomous Prefecture of Yunnan Province, and Yulin City of Shaanxi Province, accounting for 43.4% of the total area of the country. The southeast part includes all remaining parts of China, accounting for 56.6% of the total area of the country [39]. The southeast and northwest parts contain 456 (94.6%) and 26 (5.4%) buildings, respectively, indicating a ratio (94.6:5.4) that is approximately equal to the population ratio.

Six of the seven driving factors analyzed in this study have a significant influence on the spatial distribution of Ting, Tai, Lou, and Ge buildings (

Table 3. Statistical results (q-statistics and p-values) of factors driving the spatial distribution of Ting, Tai, Lou, and Ge buildings.

	Factor	China		Southeast		Northwest
		q	p	q	p	q	p
Natural	DEM	0.022	0.088 *	0.049	0.011 **	0.053	0.475
	NDVI	0.019	0.150	0.043	0.030 **	0.044	0.638
	Waterway	0.052	0.003 ***	0.099	0.003 ***	0.063	0.394
Socio	Road	0.131	0.000 ***	0.145	0.000 ***	0.065	0.438
economic	Railway	0.108	0.012 **	0.100	0.060 *	0.035	0.723
	Population	0.156	0.000 ***	0.104	0.007 ***	0.104	0.202
	CO2	0.051	0.010 ***	0.049	0.036 **	0.033	0.754

Note: *, **, and *** indicate a significance level of 10%, 5%, and 1%; p-value of CO₂ emissions in China before rounding to three decimal places was 0.009981.

). Regarding natural factors, waterway length and topography explain 5.2% (p < 0.01) and 2.2% (p < 0.1) of the heterogeneity, respectively. Conversely, the NDVI does not significantly influence the spatial distribution of buildings. Regarding socioeconomic driving factors, population, road length, and CO₂ emissions account for 15.6% (p < 0.01), 13.1% (p < 0.01), and 5.1% (p < 0.01) of the heterogeneity, respectively. The q-statistic for railway length is 10.8% (p < 0.05). Thus, socioeconomic factors have a greater influence than natural factors. All factors have a significant influence in Southeast China, whereas all factors have non-significant influences in Northwest China. This may be because the northwest region contains very few buildings. The NDVI shows different influences in Southeast China and China as a whole, exhibiting a significant influence on the distribution of buildings in Southeast China. This could be because these buildings were built for various purposes, and approximately 94% of people live in Southeast China, where Ting, Tai, Lou, and Ge buildings are typically constructed in landscape gardens for sightseeing purposes.

4.4. Rating and Heat Scores

Table 4. Number of provinces containing Ting, Tai, Lou, and Ge buildings with reviews, rating scores, and heat scores.

Number of Provinces	Ting	Tai	Lou	Ge
Total	14	18	27	22
With review	6	9	17	8
With rating scores	6	9	17	8
With heat scores	5	8	17	8

shows the number of provinces containing Ting, Tai, Lou, and Ge buildings, as well as the number of provinces containing buildings with available rating scores, heat scores, and user reviews. The numbers of rating scores and reviews are equal. The lower number of provinces with heat scores reflects the impact of very few user visits, which can result in there being no heat score. Lou buildings have rating and heat score data for the highest number of provinces, followed by Tai, Ge, and Ting buildings. However, more provinces contain Ge buildings than Tai buildings, indicating that Ge buildings are relatively less publicized.

Figure 7 shows the average rating scores for Ting, Tai, Lou, and Ge buildings in each province of China. Ting buildings with rating scores are predominantly distributed in southern China. Tai and Ge buildings show similar distributions. Lou buildings with rating scores are distributed in a north–south pattern. The highest average rating scores for Ting, Tai, Lou, and Ge buildings are 4.80, 5.00, 4.70, and 4.55 in Hubei, Shandong and Fujian, Heilongjiang, and Jiangsu and Henan, respectively, whereas the lowest average rating scores are 3.75, 4.00, 3.95, and 3.70, in Anhui, Shanxi, Hebei, and Jiangxi, respectively. The highest average heat scores for Ting, Tai, Lou, and Ge buildings are 3.50, 4.80, 5.20, and 3.65 in Yunnan, Hubei, Guizhou, and Guizhou, respectively, whereas the lowest average heat scores are 1.23, 1.50, 1.93, and 1.60 in Zhejiang, Shanxi, Guangdong, and Jiangxi, respectively (Figure 8). For China as a whole, Lou buildings have the highest average rating score and heat score of 4.40 and 3.25, respectively, whereas all buildings combined have an average rating score and heat score of 4.35 (out of 5.00) and 3.00 (out of 10.00), respectively (

Table 5. Average rating and heat scores for Ting, Tai, Lou, and Ge buildings across China.

	Rating Score	Heat Score
All	4.35	3.00
Ting	4.23	2.35
Tai	4.32	2.76
Lou	4.40	3.25
Ge	4.27	2.71

). Thus, compared to the Imperial Palace in Beijing (rating score and heat score of 4.8 and 10, respectively), Ting, Tai, Lou, and Ge buildings are still relatively attractive to tourists but receive much fewer visitors, likely because they are less well publicized. Accordingly, the government should raise public awareness and improve tourist attractiveness for buildings of valuable traditional culture, especially in provinces with the lowest heat scores for Ting, Tai, Lou, and Ge buildings. Such endeavors will promote the dissemination of traditional culture and economic growth. Additionally, provinces with the lowest average rating scores for Ting, Tai, Lou, and Ge buildings should aim to improve their service capabilities.

4.5. Sentiment Analysis

4.5.1. Kernel Density Results

Table 6. Number of Ting, Tai, Lou, and Ge buildings with positive and negative reviews.

	Ting	Tai	Lou	Ge	Total
Positive	7	12	48	11	78
Negative	2	3	3	3	11
Percent (%)	77.78	80.00	94.12	78.57	87.64

shows the number of Ting, Tai, Lou, and Ge buildings with positive and negative reviews based on sentiment analysis. Lou buildings have the most positive reviews (94.12%), followed by Tai buildings (80.00%), then Ting and Ge buildings, which have similar proportions of positive reviews. These results indicate that Lou buildings provide the best services, and that tourism services for the other three types of building should be improved to attract more tourists. According to the kernel density results of sentiment scores for Ting, Tai, Lou, and Ge buildings (Figure 9 and Figure 10), Ting buildings have only one location where the highest sentiment scores are recorded, which is concentrated around Zhejiang, whereas Tai, Lou, and Ge buildings have multiple locations where the highest and second-highest sentiment scores are recorded, all distributed in eastern China. The distribution of kernels for all buildings combined is similar to that of Lou buildings but reflects a higher density of sentiment scores.

4.5.2. Word Cloud Results

Table 7 demonstrates the four main types of word appearing in user reviews, along with their corresponding frequency for Ting, Tai, Lou, and Ge buildings. Figure 11 presents the word clouds generated based on the frequency of words used in reviews of Ting, Tai, Lou, and Ge buildings. The word cloud for Ting buildings indicates the popularity of historical stories in the user reviews (Figure 11a), with high-frequency words including Yu Zhou, who was an important general of Eastern Wu during the Three Kingdoms period, and Yanshui Ting, a pavilion thought to have been the location of Yu Zhou’s Dianjiang Terrace, where officers were called to muster and given responsibilities in ancient times. In addition to historical figures and events, words related to tourism are frequently mentioned, such as scenic spot, scenery, admission ticket, and cost-effectiveness. Furthermore, history, culture, and building are representative culture-related words. Finally, Jiujiang city, Chuzhou, Wenzhou, Anhui Province, and regions south of the Yangtze River are the provinces or cities mentioned in the reviews.

Table 7. Major word types and their corresponding frequency for Ting, Tai, Lou, and Ge sites.

Type	Site	Word (Frequency)
Historical story	Ting	Yu Zhou (周瑜, 43), Three Kingdoms period (三国, 16); Yanshui Ting (烟水亭, 81), Dianjiang Terrace (点将台, 30), Juyi Bai (白居易, 12), Xiu Ouyang (欧阳修, frequency = 51), Northern Song Dynasty (北宋, 15), famous essay (醉翁亭记, 32), Zuiweng Ting (醉翁亭, 94), Zuiweng Ting (醉翁亭), Chuzhou (滁州, 36), Liangya Mountain (琅琊山, 45)
	Tai	Boya Yu (俞伯牙, 43; 伯牙, 55), Zhong Ziqi (钟子期, 41; 子期, 25), a kindred spirit with Boya Yu through music (知音, 82), ‘High mountains and flowing water’ (高山流水, 62), Guqin Tai (古琴台, 89; 伯牙台, 3), Song Dynasty (北宋, 33), Huanghe Lou (黄鹤楼, 29), Qingchuan Ge (晴川阁, 27)
	Lou	none
	Ge	none
Tourism	Ting	scenic spot (景点, 42; 景区, 46), scenery (景色, 26), admission ticket (门票, 18), cost-effectiveness (性价比, 8)
	Tai	scenic spot (景点, 68;景区, 44), park (公园, 47), admission ticket (门票, 55), cost-effectiveness (性价比, 14)
	Lou	scenic spot (景点, 195; 景区, 158), scenery (景色, f = 141), landmarks (地标, 40; 地标性建筑, 30; 地标建筑, 27), tourist (游客, 72), admission ticket (门票, 80), free (免费, 72), cost-effectiveness (性价比, 70), visit (参观, 83)
	Ge	scenic spot (景点, 54; 景区, 18), scenery (景色, 33), landmark (地标, 19), admission ticket (门票, 26), free (免费, 39), and cost-effectiveness (性价比, 10)
Culture	Ting	history (历史, 39), culture (文化, 15), building (建筑, 14)
	Tai	astronomy (天文, 63), instrument (仪器, 81), ancient observatories (古观象台, 57), place (地方, 81), history (历史, 76), building (建筑, 57), story (故事, f = 36), culture (文化, f = 35)
	Lou	history (历史, 405), building (建筑, 368), gate tower(城楼, 253), place (地方, 237), ancient city (古城, 171), historic building (古建筑, 99), Key Cultural Relics Protection Units (重点文物保护单位, 46)
	Ge	building (建筑, 202), place (地方, 84), history (历史, 83), historic building (古建筑, 30), ancient city (古城, 29), historic site (古迹, 19), Cultural Relics Protection Unit (重点文物保护单位, 17; 文物保护单位, 16), culture (文化, 20)
Cities/ provinces	Ting	Jiujiang city (九江, 51; 九江市, 21), Chuzhou (滁州, 36, 滁州市, 8), Wenzhou (温州, 6), Anhui Province (安徽省, 7), regions south of the Yangtze River (江南, 11)
	Tai	Wuhan (武汉, 56; 武汉市, 23), Gaoyou (高邮, 49), Beijing (北京, 44; 北京市, 17), Kaifeng (开封, 23), Hubei (湖北省, 17), Hanyang district (汉阳区, 15; 汉阳, 14)
	Lou	Beijing (北京, 112; 北京市, 7), Ningbo (宁波, 77; 宁波市, 25), Xi’an (西安, 73), Guiyang (贵阳, 66), Liuzhou (柳州, 48), the Old Town of Phoenix (凤凰古城, 43), Xuhuan District (宣化, f = 139), Jinhua (金华, 25), Shunde (顺德, f = 24)
	Ge	Yangzhou (扬州, 113; 扬州市, 21), Qingdao (青岛, 83), Baoding (保定, 66; 保定市, 20), Guiyang (贵阳, 52)

Table 7. Major word types and their corresponding frequency for Ting, Tai, Lou, and Ge sites.

Type	Site	Word (Frequency)
Historical story	Ting	Yu Zhou (周瑜, 43), Three Kingdoms period (三国, 16); Yanshui Ting (烟水亭, 81), Dianjiang Terrace (点将台, 30), Juyi Bai (白居易, 12), Xiu Ouyang (欧阳修, frequency = 51), Northern Song Dynasty (北宋, 15), famous essay (醉翁亭记, 32), Zuiweng Ting (醉翁亭, 94), Zuiweng Ting (醉翁亭), Chuzhou (滁州, 36), Liangya Mountain (琅琊山, 45)
	Tai	Boya Yu (俞伯牙, 43; 伯牙, 55), Zhong Ziqi (钟子期, 41; 子期, 25), a kindred spirit with Boya Yu through music (知音, 82), ‘High mountains and flowing water’ (高山流水, 62), Guqin Tai (古琴台, 89; 伯牙台, 3), Song Dynasty (北宋, 33), Huanghe Lou (黄鹤楼, 29), Qingchuan Ge (晴川阁, 27)
	Lou	none
	Ge	none
Tourism	Ting	scenic spot (景点, 42; 景区, 46), scenery (景色, 26), admission ticket (门票, 18), cost-effectiveness (性价比, 8)
	Tai	scenic spot (景点, 68;景区, 44), park (公园, 47), admission ticket (门票, 55), cost-effectiveness (性价比, 14)
	Lou	scenic spot (景点, 195; 景区, 158), scenery (景色, f = 141), landmarks (地标, 40; 地标性建筑, 30; 地标建筑, 27), tourist (游客, 72), admission ticket (门票, 80), free (免费, 72), cost-effectiveness (性价比, 70), visit (参观, 83)
	Ge	scenic spot (景点, 54; 景区, 18), scenery (景色, 33), landmark (地标, 19), admission ticket (门票, 26), free (免费, 39), and cost-effectiveness (性价比, 10)
Culture	Ting	history (历史, 39), culture (文化, 15), building (建筑, 14)
	Tai	astronomy (天文, 63), instrument (仪器, 81), ancient observatories (古观象台, 57), place (地方, 81), history (历史, 76), building (建筑, 57), story (故事, f = 36), culture (文化, f = 35)
	Lou	history (历史, 405), building (建筑, 368), gate tower(城楼, 253), place (地方, 237), ancient city (古城, 171), historic building (古建筑, 99), Key Cultural Relics Protection Units (重点文物保护单位, 46)
	Ge	building (建筑, 202), place (地方, 84), history (历史, 83), historic building (古建筑, 30), ancient city (古城, 29), historic site (古迹, 19), Cultural Relics Protection Unit (重点文物保护单位, 17; 文物保护单位, 16), culture (文化, 20)
Cities/ provinces	Ting	Jiujiang city (九江, 51; 九江市, 21), Chuzhou (滁州, 36, 滁州市, 8), Wenzhou (温州, 6), Anhui Province (安徽省, 7), regions south of the Yangtze River (江南, 11)
	Tai	Wuhan (武汉, 56; 武汉市, 23), Gaoyou (高邮, 49), Beijing (北京, 44; 北京市, 17), Kaifeng (开封, 23), Hubei (湖北省, 17), Hanyang district (汉阳区, 15; 汉阳, 14)
	Lou	Beijing (北京, 112; 北京市, 7), Ningbo (宁波, 77; 宁波市, 25), Xi’an (西安, 73), Guiyang (贵阳, 66), Liuzhou (柳州, 48), the Old Town of Phoenix (凤凰古城, 43), Xuhuan District (宣化, f = 139), Jinhua (金华, 25), Shunde (顺德, f = 24)
	Ge	Yangzhou (扬州, 113; 扬州市, 21), Qingdao (青岛, 83), Baoding (保定, 66; 保定市, 20), Guiyang (贵阳, 52)

Figure 11. Word clouds showing the frequency of words used in reviews of (a) Ting, (b) Tai, (c) Lou, and (d) Ge buildings.

Figure 11. Word clouds showing the frequency of words used in reviews of (a) Ting, (b) Tai, (c) Lou, and (d) Ge buildings.

The word cloud for Tai buildings (Figure 11b) includes similar types of high-frequency words related to historical stories, tourism, culture, and cities and provinces. Notably, the word clouds for Lou buildings (Figure 11c) and Ge buildings (Figure 11d) do not include historical stories or figures. In general, Ting and Tai buildings show similar word clouds, as do Lou and Ge buildings, with only the former including historical stories. Conversely, landmark is a high-frequency word only in the reviews of Lou and Ge buildings. Although landmark appears in the word cloud for Ting buildings, the frequency is low (3).

These findings provide important insights for promoting tourism at key Provincial Cultural Relic Protection Units of China. Specifically, our results suggest that the historical stories of Lou and Ge buildings should be highlighted. Intangible cultural heritage, which is a crucial cultural resource for the high-quality growth of the tourism industry [17], can be attached to tangible cultural heritage. In this way, visitors also learn cultural stories while sightseeing [40], ensuring memorable and impactful experiences [41], which can strengthen cultural identity, confidence [42] and a sense of belonging [43]. Making a memorable impression is crucial for expanding the tourism market of heritage buildings [28]. In addition, Ting and Tai buildings should be promoted by shaping these buildings into landmarks, which form one of the five elements (i.e., landmarks, nodes, paths, edges, and districts) comprising the image of a city/tourism destination [44,45]. Finally, the reviews of Ting, Tai, Lou, and Ge buildings indicate several cities and provinces that are well-known locations with close links to these buildings. Therefore, links with the cities/provinces not mentioned in these reviews should be strengthened by creating city tourism cards or brands.

5. Conclusions and Future Work

Ting, Tai, Lou, and Ge are types of ancient buildings that represent traditional Chinese architecture and culture. However, research aimed at conserving, inheriting, and rejuvenating these buildings is limited, despite their status as Provincial Cultural Relic Protection Units of China. Therefore, the aim of this study was to reveal the spatial distribution of Ting, Tai, Lou, and Ge buildings across China, as well as the factors driving differences in their spatial distribution. Tourist experiences and building popularity were also explored through the spatial analysis and sentiment analysis of Ctrip data.

The key findings of this study are as follows:

(1)

The ratio of Ting, Tai, Lou, and Ge buildings in Southeast China to that in Northwest China was approximately the same as that of the population (94:6).

(2)

Geographic detector analysis revealed that six of the seven natural and socioeconomic factors (topography, waterways, roads, railways, population, and CO₂ emissions) had a significant influence on the spatial heterogeneity of these cultural heritage sites in China, with socioeconomic factors, particularly population, having a greater influence on building spatial distributions. All seven factors (including the NDVI) were significant in Southeast China, whereas all factors were non-significant in Northwest China, which may be explained by the small number of buildings in the latter region.

(3)

The average rating and heat scores for Ting, Tai, Lou, and Ge buildings reflected an imbalance between service quality and popularity. The service quality was acceptable whereas the popularity should be improved.

(4)

Four main types of words were used with high frequency in the tourism reviews: historical stories, tourism, culture, and cities/provinces. Ting and Tai buildings showed similar word clouds, as did Lou and Ge buildings, with only the former including historical stories. Conversely, landmark was a high-frequency word only in the reviews of Lou and Ge buildings.

In summary, this study revealed the spatio-temporal patterns of Ting, Tai, Lou, and Ge buildings, as well as the factors driving spatial heterogeneity, and explored tourist perceptions of these buildings through sentiment analysis. Moreover, suggestions were proposed for improving tourism strategies at Ting, Tai, Lou, and Ge buildings and rejuvenating ancient Chinese architecture. However, several limitations exist. First, the traditional cultural aspects of these buildings, such as the esthetics, architecture, and social values, were not explored. Second, additional driving factors, such as government investment, land cover, and climate, were not considered. Finally, only the Ctrip website was used to gather review data, neglecting other popular tourism websites in China such as Mafengwo. Nevertheless, this study contributes to the conservation, inheritance, and rejuvenation of Chinese cultural heritage through the integration of culture and tourism.