Comparative Study of Park Evaluation Based on Text Analysis of Social Media: A Case Study of 50 Popular Parks in Beijing

Abstract

With China’s urban renewal, parks have developed into significant green recreational areas in cities. This paper analyzed social media texts and compared the evaluation outcomes of the 50 most popular urban parks in Beijing from various perspectives, such as the characteristics of various groups of people, park types, and the spatial and temporal distribution characteristics of recreational activities. The importance–performance analysis method was used to analyze the main factors affecting visitors’ satisfaction with parks. The research found the following: (1) Positive evaluation of parks was related to environmental construction, event organization, etc., and negative evaluations focused on ticket supply, consumer spending, etc. (2) Visitors of different genders and from different regions focused on different aspects of parks. (3) In terms of traffic accessibility, historical and cultural display, parent–child activity organization, and ecological environment experience, people had diverse demands from various types of parks. (4) People were more likely to visit parks located within the range of all green belts in springs and parks located in the second green isolation belt in the fall. (5) The number of non-holiday reviews of parks was higher than that of holiday reviews. (6) Managers could improve visitor satisfaction by improving the infrastructure and management of parks.

1. Introduction

Urban development in China is currently shifting from an expansion type to a connotation-promotion kind. Urban renewal is an important part of China’s land and space planning and management, backed by the revitalization of stock land and the optimization of spatial structure. For the first time, in the government work report in 2021, National People’s Congress and the Chinese People’s Political Consultative Conference included the content of urban renewal. The Fourteenth Five-Year (2021–2025) Plan for National Economic and Social Development of the People’s Republic of China and the Vision for 2035 also clearly proposed the implementation of urban renewal actions to improve the urban spatial structure, to enhance the environment quality, and to increase the happiness of people. The urban renewal actions specifically proposed the sustainable development goals of building livable cities, green cities, resilient cities, smart cities, and humanistic cities [1]. As the natural ecosystem and the green infrastructure formed by the urban green space are the life support systems on which the sustainable development of the city depends, their renewal must be shifted from a crude to a refined planning mode, which means a shift from the previous focus on the renewal of gray infrastructure, such as urban corridors and roads, to the emphasis on the renewal of green infrastructure, such as urban parks and green corridors. This is one of the important directions of urban renewal [2]. In many successful urban renewal cases in the past, such as Paley Park [3] and the Fuzhou Urban Greenway, an ecologically livable urban environment was often reflected in the renewal of the urban green infrastructure [4].

Urban parks provide citizens with spaces nearby for leisure activity, encouraging people to “open the window to see the green” and “go out to see the garden,” enhancing the quality of living and working environment, and thus increasing people’s happiness. According to official data from the Beijing Landscaping Publicity Center, in the past three years, only during five statutory holidays (May Day, Mid-Autumn Festival, National Day, Dragon Boat Festival, and Spring Festival) did the cumulative number of visitors received by Beijing’s park scenic spots reach extremely high figures. The numbers were 32.06 million in 2019, 20.03 million in 2020, and 21.47 million in 2021. In 2020 and 2021, the flow of people had decreased due to the impact of the epidemic. Therefore, urban parks have become an inseparable part of people’s leisure life. Beijing is one of the first batches of urban renewal pilot cities in China. Optimizing the functions of urban parks has become a crucial component of their urban renewal policy, and the public’s opinions on how parks are used must be taken into consideration while working on urban renewal projects.

The majority of the traditional current studies evaluating the role of parks as green spaces have been built on the survey analysis of the data provided by questionnaires on individual parks. These studies had limitations such as limited content categories, small data sample size, and insufficient collection time, which affected the universality and accuracy of research results to a certain extent [5,6,7]. Because of this, it has been uncommon to compare and assess the perspectives of diverse users on how different park services function based on the two dimensions of time and space.

As network information technology has grown in popularity, so have the avenues for citizens to participate in the public evaluation of urban parks. An increasing number of scientific studies have started paying attention to the humanistic feelings behind the data and to interpret cities from the bottom-up ideological perspective [8,9,10]. Today, big data research on urban parks and green spaces uses cell phone signaling, satellite positioning, social media, and other platform data for analysis. Niu et al. [11,12,13] used mobile phone signaling data to extract population distribution in a timely manner through the kernel density method, a new idea for the study of urban spatial structure. Li et al. [14,15,16,17] studied the features of people’s travel changes, using satellite positioning and GIS, and supplemented new methods to further improve the experience of urban green space environment. Sinclair et al. [18,19] gathered and used images with geographic location information in combination with the site’s real situation and the public’s input and used this knowledge to drive the overall urban design.

Social media data are among the multi-source urban data. Social media is people-centered and disseminates information about daily life and personal experiences. Many scholars have recognized social media as a tool for public participation. It allows us to re-examine our understanding of urban parks [20,21,22,23]. Use of online social media as a platform for research and analysis could provide a way for the public to participate in the optimization of urban park functions, in addition to helping create an urban environment that meets people’s requirements as well as further improves people’s happiness, which is of great significance to the function optimization of urban parks against the background of urban renewal.

The commonly used social network data in China are from social media websites or apps such as Weibo, Xiaohongshu (https://www.xiaohongshu.com/ (accessed on 29 December 2021)), Mafengwo (https://www.mafengwo.cn/ (accessed on 29 December 2021)), and Dianping (https://www.dianping.com/ (accessed on 29 December 2021)), which capture visitors’ check-in data and subjective review data to obtain information about the their experience. Among them, Weibo and Xiaohongshu websites have a high coverage of Internet users, but the park reviews posted by users are mostly impromptu and short, with little effective content, and are mostly used to study the check-in behavior of users at specific locations. Fujisaka et al. [24] used Weibo data to compare and study the behaviors of different users traveling in different regions. Zhou et al. [25] used the geographic check-in data provided by Weibo to compare the number of tourists in different types of parks in Beijing and analyzed their influencing factors. The comments posted by users on the Mafengwo website are rich in information, but the website is not popular, it has few registered users, and the sample size available is small, which may lead to large errors in the research results, so almost no scholars have used user reviews from the Mafengwo website for scientific research.

Dianping (https://www.dianping.com/ (accessed on 29 December 2021)) is China’s leading urban lifestyle consumer platform and an independent third-party consumer review site, with extensive user data coverage, rich review content, and full expression of emotions, providing more research advantages than other online platforms. All information on the website comes from user registration and comments. Except for classifying and integrating information, the platform itself does not edit the content, and all users can share and use it without influencing each other [26]. On the platform, users rate and score an attraction or product on the basis of their personal consumption and experience, and the platform aggregates this information in the fastest time to provide reference for other users subsequently [27]. By collecting many valid reviews of the parks that people have visited from the platform, it is possible to analyze and compare the results of people’s evaluation of the parks.

On the basis of this background and theory, via social media data text analysis, this paper compared the evaluation results of the 50 most popular urban parks in Beijing from multiple perspectives. The following three aspects would be identified:

(1)

The difference in star ratings for all parks and the difference between their positive and negative evaluations at different time periods;

(2)

The influence of different population characteristics, different park types, different visiting time periods, and different park locations on park evaluation;

(3)

The factors that influence visitors’ recreational satisfaction and sensory satisfaction based on IPA.

2. Literature Review

2.1. Text Word Frequency Analysis

Traditional text analysis methods are mostly used to analyze the perceptions and preferences of visitors to a destination at a specific time, with the goal being to benefit managers. Wallace and Reed [28] used traditional questionnaires to obtain information about visitor characteristics, preferences, and satisfaction to help park managers identify how park values and resources are being impacted. Mameno et al. [29] quantified the aesthetic and economic value of Daisetsuzan National Park by analyzing data from 445 respondents. Chiesi et al. [30] investigated human perception and use of small green spaces in a dense western city (Florence, Italy) by distributing questionnaires and by conducting interviews.

Social media text analysis has been mostly used in diverse research on recreational experience. A large amount of the latest online evaluation data has helped to identify people’s true feelings about the destination and helped managers to clearly plan strategies and tactics from various perspectives. Through text analysis, Lin et al. [31] proposed improvement strategies for Haizhu National Wetland Park in terms of infrastructure, management services, scenic spot publicity, and ecological environmental protection. Wan et al. [32] investigated users’ preferences and values related to urban parks by analyzing social media data and obtained important information for formulating social marketing strategies. Jin et al. [33] used social media data and emotion analysis to quantify and compare the positive emotions of visitors to different types urban parks and proposed suggestions for park planning and management. Tenkanen et al. [34] used data from Instagram, Twitter, and Flickr to systematically assess the popularity of and number of visitors to several national parks in 2014. Sinclair et al. [35] used the Flickr website to predict visitor sources and recreation patterns in German national parks.

2.2. Importance–Performance Analysis

As a diagnostic model biased toward qualitative research, importance–performance analysis (IPA) can provide managers with corresponding strategies and recommendations for the sustainable development of the landscape. IPA was first proposed by Martilla and James to analyze product attributes [36]. Due to its simplicity, intuition, and ease of interpretation, IPA has been frequently introduced into the field of landscape architecture in recent years. Many scholars have used the IPA method to evaluate the importance of and satisfaction with each indicator of green space, so as to study the recreation demand of park visitors and the supply–demand relationship of the cultural ecosystem. Yu et al. [37] constructed an IPA model to reflect visitors’ basic attitudes, entertainment sense, and satisfaction evaluation, as well as their preferences regarding the future development of community parks. Fan et al. [38] coded and rated the recreational factors of Fuzhou West Lake Park by the IPA method. Wang et al. [39] analyzed the supply and demand characteristics of cultural ecosystem services in the waterfront space through the IPA model and proposed strategies to optimize the supply of related services.

It is essential to establish a complete park evaluation system via the IPA approach. Jahani et al. [40] modeled some aesthetic preferences and spiritual recovery values of urban parks by extracting 11 landscape features to compare the aesthetic and spiritual recovery potential of urban parks. Liu et al. [41] set up a recreational evaluation system with traffic, facilities, landscape, activities, culture, and management as the main evaluation items by extracting the high-frequency words from the network to evaluate waterfront public spaces and obtained the main factors affecting the satisfaction of visitors with public space recreation. Zheng et al. [42] analyzed the terrestrial perceptual elements in the park elements; developed a perceptual evaluation system with visual, auditory, olfactory, tactile, and gustatory senses as the main evaluation items for the four human senses; and evaluated the urban park from the perspective of scenic sense ecology. This study will comprehensively refer to relevant literature and combine statistics on specific high-frequency words in order to form a suitable park recreational evaluation system and sensory perceptual evaluation system for further analysis.

Since social media data are generally collected all at once by a web crawler written in Java language, most of the domestic and foreign studies on web text data of park green spaces focus on the analysis of text content within a single time period and the comparative analysis in the time dimension is relatively weak. For example, Munawir et al. [43] only explored the overall visitor experience of Bandung theme park by obtaining user reviews from Google Maps. Sim et al. [44] only studied social media users’ overall feelings about the High Line over time by means of text mining. However, Wang et al. [45] analyzed the change in people’s demand for park green space by comparing the differences in people’s attention to park elements over three years, which effectively broadened the temporal scope of the park evaluation study and had a strong guiding significance for this study. Therefore, currently, there are not many comparative studies on the application of text analysis to evaluate urban parks in the dimensions of time and space. The present study will take reviews from different years and different months separately for statistical processing to strengthen, highlight, and explore the differences in park reviews with changing months and years and in terms of time, which will expand and supplement the current information and have theoretical value and practical significance for the functional improvement of park green areas in urban renewal.

3. Materials and Methods

In this study, 50 highly popular parks in Beijing were used as the research objects. We (1) collected information garnered from users’ comments on each park in each time period on social media; (2) used text analysis to obtain the results of differences in park evaluation texts by gender, geographic population, category, time, and district and their reasons; and (3) explored the main factors affecting satisfaction with the recreation offered by the parks and sensory perceptual satisfaction using the importance–performance analysis (IPA) method. Ultimately, this study would help managers to identify the direction to optimize urban parks in the background of urban renewal.

3.1. The Research Object

Beijing, one of the first batches of urban renewal pilot cities in China, served as the study’s research location. To obtain comparative evaluation results of the parks from different perspectives, the public ratings and comment texts related to the top 50 highly popular Beijing city parks (

Table 1. List of 50 highly popular parks in Beijing.

Name	District	Ticket Prices	Category	Position	Score
Olympic Forest Park	Chaoyang District	0	Forest park	In the first green isolation belt	4.82
Zhongwu Park	Haidian District	0	Comprehensive park	In the first green isolation belt	4.75
Jingshan park	Xicheng District	2	Humanities park	Within the range of all green belts	4.69
Yizhuang New Town Riverside Forest Park	Daxing District	0	Forest park	In the second green isolation belt	4.68
Dongjiao Forest Park	Tongzhou District	0	Forest park	In the second green isolation belt	4.67
Beijing Wenyu River Park	Chaoyang District	0	Comprehensive park	In the second green isolation belt	4.65
Grand Canal Forest Park	Tongzhou District	0	Forest park	In the ring of forest wetland parks surrounding the capital	4.65
Xishan National Forest Park	Haidian District	10	Forest park	In the second green isolation belt	4.64
Temple of Heaven Park	Dongcheng Area	12.5	Humanities park	Within the range of all green belts	4.63
Guangyanggu Urban Forest Park	Xicheng District	0	Forest park	Within the range of all green belts	4.62
Zhongshan Park	Dongcheng Area	3	Humanities park	Within the range of all green belts	4.61
Linglong Park	Haidian District	0	Comprehensive park	Within the range of all green belts	4.59
Beihai Park	Xicheng District	10	Humanities park	Within the range of all green belts	4.59
Nanhaizi Park	Daxing District	0	Comprehensive park	In the second green isolation belt	4.57
Shougang Park	Shijingshan District	0	Humanities park	In the second green isolation belt	4.56
Xuanwu Art Garden	Xicheng District	0	Comprehensive park	Within the range of all green belts	4.55
Changying Park	Chaoyang District	0	Comprehensive park	In the first green isolation belt	4.54
Urban Green Heart Forest Park	Tongzhou District	0	Forest park	In the ring of forest wetland parks surrounding the capital	4.54
Beigong National Forest Park	Fengtai District	7.5	Forest park	In the second green isolation belt	4.54
Niukouyu Wetland Park	Fangshan District	0	Wetland park	In the ring of forest wetland parks surrounding the capital	4.53
Changping New Town Riverside Forest Park	Changping District	0	Forest park	In the ring of forest wetland parks surrounding the capital	4.51
Tongzhou Canal Park	Tongzhou District	0	Comprehensive park	In the ring of forest wetland parks surrounding the capital	4.5
Haidian Park	Haidian District	0	Comprehensive park	In the first green isolation belt	4.49
Yuyuantan Park	Haidian District	2	Comprehensive park	Within the range of all green belts	4.48
Majiawan Wetland Park	Chaoyang District	0	Wetland park	In the second green isolation belt	4.48
Longtan park	Dongcheng Area	2	Comprehensive park	Within the range of all green belts	4.48
Rending Lake Park	Xicheng District	0	Comprehensive park	Within the range of all green belts	4.47
Lotus Pond Park	Fengtai District	2	Comprehensive park	In the first green isolation belt	4.47
Baiwangshan Forest Park	Haidian District	6	Forest park	In the second green isolation belt	4.46
Red Scarf Park	Chaoyang District	0	Comprehensive park	Within the range of all green belts	4.46
Taoranting Park	Xicheng District	2	Comprehensive park	Within the range of all green belts	4.46
Liuyin Park	Dongcheng Area	0	Comprehensive park	Within the range of all green belts	4.45
Beijing Winter Olympic Park	Shijingshan District	0	Comprehensive park	In the second green isolation belt	4.45
Beijing Expo Park	Yanqing District	50	Comprehensive park	In the ring of forest wetland parks surrounding the capital	4.45
Chaoyang Park	Chaoyang District	5	Comprehensive park	In the first green isolation belt	4.43
Ditan Park	Dongcheng Area	2	Humanities park	Within the range of all green belts	4.42
Fragrant Hills Park	Haidian District	7.5	Humanities park	In the second green isolation belt	4.41
Dongxiaokou Forest Park	Changping District	0	Forest park	In the first green isolation belt	4.4
Yeyahu National Wetland Park	Yanqing District	45	Wetland park	In the ring of forest wetland parks surrounding the capital	4.39
The Black Bamboo Park	Haidian District	0	Comprehensive park	Within the range of all green belts	4.38
Green Dam Park	Fengtai District	0	Comprehensive park	In the second green isolation belt	4.38
Daoxiang Lake Natural Wetland Park	Haidian District	0	Wetland park	In the second green isolation belt	4.37
Beijing Grand View Garden	Xicheng District	40	Humanities park	Within the range of all green belts	4.34
Cuihu Wetland Park	Haidian District	0	Wetland park	In the second green isolation belt	4.29
Moon Temple Park	Xicheng District	1	Humanities park	Within the range of all green belts	4.29
Qingnianhu Park	Dongcheng Area	0	Comprehensive park	Within the range of all green belts	4.29
Yongding River Leisure Forest Park	Shijingshan District	0	Forest park	In the second green isolation belt	4.27
Sun Palace Park	Chaoyang District	0	Comprehensive park	In the first green isolation belt	4.26
Qinglonghu Park	Fengtai District	30	Comprehensive park	In the ring of forest wetland parks surrounding the capital	4.22
Hanshiqiao Wetland Park	Shunyi District	0	Wetland park	In the ring of forest wetland parks surrounding the capital	3.88

The list of parks is from the Beijing “surrounding tour” channel of Dianping.com: https://www.dianping.com/beijing/ch35 (accessed on 28 December 2021).

) with the highest number of reviews sorted on the Dianping website were analyzed. Combined with the classification standard of urban green space in China, we proposed a set of park typologies based on park attributes specifically for this study, the 50 selected parks present four categories according to their attributes: comprehensive parks (23), humanistic parks (9), forest parks (12), and wetland parks (6). Comprehensive parks are green areas with rich contents, suitable for various outdoor activities, with perfect recreation and supporting management and service facilities. Humanistic parks are green areas focusing on specific history and culture and with recreation and leisure functions. Forest parks are forest green areas of a certain scale, with beautiful natural scenery and available for people’s recreation. Wetland parks are green areas with a good wetland ecological environment, and diverse wetland parks are green areas based on a good wetland ecological environment and diverse wetland landscape resources, with multiple functions and recreation and service facilities. The 50 selected parks were divided into four categories in accordance with the Construction and Development Plan of Greening Isolated Areas in the 14th Five-Year Period of Beijing: parks within the range of all green belts (18), parks in the first green isolation belt (8), parks in the second green isolation belt (15), and parks in the ring of forest wetland parks surrounding the capital (9) (Figure 1). The national policy stipulates the construction scope of “the first green isolation belt” mainly from the fourth ring road to the fifth ring road in Beijing, with 50% of green open space, which is a key area to serve and guarantee the ecological security pattern of the central city. The construction scope of the “ the second green isolation belt “ is the space outside the “ the first green isolation belt “ area, extending beyond the sixth ring road, with 70% of green open space, which maintains the ecological security of the city [46]. The phrase “within the range of all green belts” refers to the area from the city center to the third ring road of Beijing, while the “ring of forest wetland parks surrounding the capital” is a large-scale ecological transition green space between Beijing and the surrounding cities. Additionally, since the urban renewal process in China is closely related to the changes in the urban investment market and the overall urban investment market in China shows a fluctuation cycle of about three years [47], to study the trend in park reviews over time in the context of urban renewal, the years from 2006 to 2021 were separated into five periods: from 2006 to 2009, from 2010 to 2012, from 2013 to 2015, from 2016 to 2018, and from 2019 to 2021. Among them, the sample size of 2006 reviews was too small and there were no extensive reviews. Therefore, to facilitate a comparative study of park reviews between the five time periods, we merged 2006 into the first time period.

3.2. Data Acquisition and Processing

The data for this study were collected from the Beijing “surrounding tour” channel of Dianping.com, and all information was obtained from users’ comments based on their real experiences and consumption. Firstly, the top 50 highly popular urban parks with the maximum number of comments were selected, among which even the park with the least number of comments had more than 700 comments. The goal was to avoid the issue of high uncertainty (chance) caused because of a small number of samples. Second, information about the park evaluation was gathered using a web crawler. This information included the park name, user ID, text content of the review, number of stars in evaluation (1 to 5 stars), evaluation time, gender and region of the evaluator, number of likes and replies, etc. Since the study only included comments made between 2006 and 2021, the problem of biased results caused by delayed comments was avoided. Moreover, an effective evaluation text with abundant information and high approval had been obtained through preliminary screening of the collected information, and finally, 309,865 comment texts with 41,169,352 words in total were acquired.

3.2.1. Web Text Word Frequency Analysis

ROSTCM6 software, developed and coded by Professor Shen Yang of Wuhan University, is the only large-scale free social computing platform in China to assist humanities and social science research. Moreover, the study quantified the texts with the help of two tools, “word frequency analysis” and “social network and semantic network analysis,” in the data analysis software ROSTCM6 and developed a high-frequency-word analysis graph and a semantic network analysis graph in the review texts of the 50 urban parks in each period. The top five high-frequency words and the number of times they were used in each period were represented by a histogram in the high-frequency-word analysis graph, and the number change trend in the top five high-frequency words in all periods was shown by a broken line chart. The semantic network analysis graph was made up of network nodes and directed line segments, enabling visual study of evaluation objects. This allowed for a deeper investigation of the variables and evolving patterns that influence park evaluation in the temporal and spatial dimensions. In addition, the evaluation texts with 4.5-star ratings or above in each time period were extracted for semantic network analysis to generate positive semantic network graphs, while the evaluation texts with 3-star ratings or less in each time period were extracted to generate negative semantic network graphs.

3.2.2. Importance–Performance Analysis (IPA)

This study summarized the recreational factors and the individual sensory factors of urban parks in relevant literature, extracted the high-frequency words of network review texts, recorded people’s concentrated concerns, combined the two to determine the evaluation factors, and formed the relevant evaluation system after coding each factor. Finally, a park recreational evaluation index system with 4 major indexes and 17 factors (

Table 2. Evaluation index system of park recreation.

Evaluation Term	Indicators	Indicator Definition
Landscape (A)	Natural ecological environment (A1)	Environmental quality, beautiful scenery, etc.
	Plant landscape (A2)	Trees, leaves, flowers, etc.
	Animal landscape (A3)	Ducks, birds, fish, squirrels, frogs, etc.
	Mountain and water landscape (A4)	Mountains, rocks, water, rivers, lakes, ponds, etc.
	Historical and cultural landscape (A5)	Culture, history, royalty, red walls, ancient pavilions, the Hall of Abstinence, circular mounds, ancient trees, etc.
Activity (B)	Humanities activities (B1)	Exhibition halls, temple fairs, sacrifices, gardening, etc.
	Country activities (B2)	Boating, camping, mountain climbing, picnics, tents, etc.
	Recreational activities (B3)	Dancing, walking, taking pictures, resting, etc.
	Fitness activities (B4)	Exercise, sports, running, cycling, gym, etc.
Infrastructure (C)	Transportation accessibility (C1)	Highway, subway, bus, driving, walking, distance, location, etc.
	Public service facilities (C2)	Parking lots, restrooms, toilets, trash cans, etc.
	Navigation signage system (C3)	Navigation, maps, explanations, etc.
	Food and beverage facilities (C4)	Catering, restaurants, kiosks, ice cream, commodities, etc.
Management (D)	Consumer spending (D1)	Ticket price, free, charge, consumption, etc.
	Services provided (D2)	Management, attitude, reservations, complaints, quality, maintenance, queuing, etc.
	Planning layout (D3)	Planning, routes, areas, buildings, spaces, etc.
	Science education (D4)	Popular science, exhibitions, learning, knowledge, etc.

) and a sensory evaluation index system with 5 major indexes and 18 factors (

Table 3. Evaluation index system of sensory perception.

Senses Term	Indicators	Indicator Definition
Vision (E)	Visual identification (E1)	Vision of special sights
	Vision of plants (E2)	Vision of trees, grass, flowers, etc.
	Vision of water (E3)	Vision of water
	Vision of animals (E4)	Vision of wild ducks, squirrels, birds, etc.
	Vision of humans (E5)	Moderate number of people and no interference
	Vision of roads (E6)	Vision of the line shape, color, etc. of the road
Hearing (F)	Sound of voice (F1)	Moderate voice
	Sound of broadcast (F2)	Sound of broadcast
	Sound of animals (F3)	Sound of birds, insects, etc.
	Sound of water (F4)	Sound of water flow
Smell (G)	Smell of air and water (G1)	Fresh air and good water quality
	Smell of plants (G2)	Smell of plants
Touch (H)	Feel of sunlight (H1)	Feel of the balance of light and shadow
	Feel of wind (H2)	Feel of wind
	Touch of roads (H3)	Comfortable roads
	Feel of water (H4)	Hydrophilic experience
	Contact with animals (H5)	No mosquito bites
Taste (L)	Food available (L1)	Food available

) were formed. The importance of each evaluation factor to visitors was measured by its occurrence frequency, and the satisfaction of visitors with that factor in terms of the five senses was expressed by the proportion of positive emotion text in all the text data. Firstly, this study conducted frequency statistics of the number of words corresponding to each indicator based on the index system. Secondly, the evaluation texts were graded by referring to the satisfaction rating method of the Likert point scale (1 star = very dissatisfied, 2 stars = dissatisfied, 3 stars = general, 4 stars = satisfied, and 5 stars = very satisfied) [1]. The evaluation texts that were given a rating of above 4 stars were regarded as texts expressing positive emotion. Then, the number of positive-emotion texts were counted according to each index and the proportion of positive-emotion texts was used to measure the degree of satisfaction of users with that specific element.

The IPA matrix took the visitor expectation (importance) as the horizontal axis, the visitor satisfaction as the vertical axis, and the overall average value as the separation point of the X–Y axis and formed an evaluation model that included areas that continue to strive to develop, areas that do not need much improvement, areas that need gradual slow improvement, and areas that need major improvement (Figure 2), which can provide guidance for the landscape planning and construction of parks.

4. Results

4.1. Comparative Evaluation of 50 Popular Parks in Beijing

From the park ratings, we can observe the variation in how the general public views these 50 popular parks, with 22 getting ratings greater than 4.50 stars and just 1 getting a rating lower than 4 stars.

To understand the elements that influence citizens’ ratings of parks and the changes in these elements over time, we generated positive semantic network illustrations (Figure 2) and negative semantic network illustrations (Figure 3). To understand the factors that affect citizens’ ratings of parks and the changes in these factors in each period, we extracted the evaluation texts with a rating of 5 stars in each period to perform high-frequency-word and semantic network analysis and developed a positive semantic network diagram (Figure 3). To create a negative semantic network map, the assessment texts from each period with a rating of 3 stars or less were retrieved (Figure 4). The statistical results show that, in the positive semantic evaluation, citizens focused all the time on two elements: ticket and environment. From 2016 to 2021, two new elements of focus were added: attitude and photography. However, in the negative semantic assessment, across all time periods, “free” and “children” were the aspects that visitors continued to focus on, while “parking lot” was a newly added element of concern from 2013 to 2021. Appropriate price and quantity of tickets, beautiful and pleasant environment, sufficient play space for children, and fully functional parking lot were clearly the elements essential for enhancing the public’s evaluation of the park.

4.2. Differences in Park Evaluation by Populations with Different Characteristics

We studied 64,373 men and 198,832 women in order to analyze how the park ratings varied by various population characteristics. The average rating by male reviewers was 4.51 stars, and that by female reviewers was 4.53 stars. Therefore, the ratings did not differ significantly. In the study period, there were 262,839 locals and 45,748 non-locals in Beijing. The average rating by local reviewers was 4.53 stars, and that by non-local reviewers was 4.52 stars. There was not much variation in star ratings among locations.

4.2.1. Differences in Park Evaluation between Different Genders

In each time period, male reviewers paid greater attention to the park’s environment, scenery, and ticket availability (Figure 5 and Figure 6). From 2019 to 2021, they gradually started paying attention to the element of traffic. In contrast, women reviewers generally paid attention to the “consumer price” of the park as well as the adaptability of children and the elderly to the park. Between 2006 and 2009, female reviewers paid greater attention to the park’s environment and tickets. From 2019 to 2021, both male and female reviewers steadily shifted their attention to convenience elements. Therefore, parking convenience was a factor of concern for both and women were more concerned with improving all-age friendliness of park services.

4.2.2. Differences in Park Evaluations by Visitors from Different Regions

The park features that 262,839 local visitors and 45,748 non-local visitors were concerned about were different (Figure 7 and Figure 8). Local visitors have always been more concerned with the ticket supply and started focusing on the parks’ all-age service and parking lots, while non-local visitors were more interested in the parks’ architecture and scenic site characteristics. From 2019 to 2021, non-local visitors gradually began to pay attention to the issue of traffic accessibility. Moreover, data analysis shows that, since local visitors accounted for nearly 86% of the collected samples, in all parks, far more local visitors than non-local visitors left reviews, but the number of reviews for Temple of Heaven Park was almost similar. Non-local visitors were seen to be more inclined to the parks with prominent cultural landscapes.

4.3. Differences in Evaluations of Different Park Types

According to their various characteristics, the chosen parks were classified into four categories: comprehensive parks (23), humanities parks (9), forest parks (12), and wetland parks (6). The average rating of all parks had risen over time, as shown by data on the average ratings (stars) of and average number of comments on various parks each year. The average rating of the forest parks had changed the most, while the average rating of the humanities parks had changed the least. The average number of comments on the humanities parks were the highest (12,646), while that on the wetland parks were the lowest (2007). Additionally, the average numbers of comments on the comprehensive parks and the forest parks were 4931 and 4838, respectively.

The change chart on high-frequency words in the evaluation of comprehensive parks (Figure 9) shows that people focused on keywords such as “cherry blossom,” “dancing,” and “convenience,” primarily including the needs of scene viewing and daily convenience activities. Among them, “cherry blossom” had steadily gained more attention, indicating that the concentrated plant landscape in the comprehensive parks at certain periods were a considerable draw for visitors, and sightseeing was the main activity of visitors to comprehensive parks. Simultaneously, the frequency of the words “convenience” and “area” continued to be high, demonstrating that visitors attached high significance to the accessibility as well as activity area of comprehensive parks.

In the evaluation of humanities parks (Figure 10), people focused more on keywords such as “attraction” and “convenience.” The focus on specific attractions, such as “Tian An Men,” “ Fragrant Hill,” and “Hall of Prayer for Good Harvests, was also high. These keywords illustrate that visitors preferred humanities parks with the strong purpose of gaining information about Beijing’s historical culture, classical gardens, and architecture. Popular special events in certain parks included “red leaf” and “temple fair.” The popularity of each keyword remained almost unchanged, indicating that visitors’ attention to humanities parks had not changed much.

People focused more on keywords such as “area,” “children,” and “entertainment” (Figure 11), demonstrating that visitors’ demands for forest parks mainly lay in the closeness of natural forest landscapes and parent–child activities. The emphasis on “parking lot” demonstrates the broad range of services offered by forest parks. Many visitors chose to drive to forest parks that were farther away from the residence for activities.

In the evaluations of wetland parks (Figure 12), the focus of people was mainly on “wetland,” “wild duck,” and “bicycle,” showing that the main attraction of wetland parks was the wetland landscape, wildlife, and cycling activities. The words “parking lot” and “hour” that were connected to the trip also received a lot of interest, indicating that some visitors opted to drive to far wetland areas. The attention to “wild duck” continued to rise every year, while the word “barbecue” gradually declined, indicating that the environmental management of the park had been strengthened and visitors’ ecological experience had been enhanced in recent years.

An analysis of high-frequency words (Figure 13) shows that words such as “convenience”, “area”, “comfortable”, and “ticket” were used the maximum to describe the four types of parks, demonstrating that visitors’ assessments of various parks were based on the parks’ accessibility, sufficient activity space, sensory experience, and cost of tourism. People visited comprehensive parks for daily activities, while they visited humanities, forest, and wetland parks for other important purposes. For instance, visitors tended to take part in cultural activities in humanities parks, while they were increasingly inclined to drive with their families on weekends and holidays to forest parks and wetland parks. The functions of these four types of parks varied in terms of the target activities of the crowd.

4.4. Differences in Park Evaluations in Different Periods

People’s happiness with the parks progressively grew between 2006 and 2021. As a result, Beijing’s parks total network score improved every year, with the average score rising steadily from 4.0 to 4.6 stars. Meanwhile, due to the widespread popularity of social networking platforms and the government’s emphasis on park construction and maintenance in the face of urban renewal in the past two years, more and more people have been using Dianping to record their personal visit experiences, resulting in a significant increase in the number of park reviews in 2019–2021, and these will continue to increase. According to the cycle theory of urban renewal and urban investment market, we selected online review texts from 2006 to 2009, from 2010 to 2012, from 2013 to 2015, from 2016 to 2018, and from 2019 to 2021 for semantic network analysis, and since the number of reviews in 2006 was too low, they were merged directly into the first cycle so we could explore in depth the elements that people are concerned about related to urban parks that have changed over time.

Statistical results (Figure 14) show that from 2006 to 2009, “environment,” “cheap,” and “air” were the factors most frequently highlighted by visitors when rating parks and “boating,” “exercise,” and “mountain climbing” were the main activities of people. From 2010 to 2012, the elements most in focus among people were “environment,” “attractions,” “facilities,” and “projects.” From 2013 to 2015, people generally focused on “environment,” “free,” “cheap,” and “building,” while “traffic,” “parking lot,” and so on also gradually received attention. From 2016 to 2018, “traffic,” “time,” “photograph,” “parking lot,” and other factors occupied people’s attention. From 2019 to 2021, the factors people focused on were more diversified. “Time,” “taking photos,” and “traffic” became the key elements of evaluation, while “attitude,” “management,” and “advice” also gradually emerged. “Environment,” “ticket,” and “free and charged” were always essential variables by which people assessed parks in these five periods. In addition, people paid increasing attention to convenience of transportation, scenic features, project facilities, management services, and other elements of the park. Simultaneously, the activities in the park gradually changed from boating and temple fairs to walking and taking photos, which represents not only the change in people’s entertainment interests but also the function of urban parks, from organizing specific cultural activities to providing daily leisure services.

4.5. Differences in Evaluation of Parks at Different Locations

To study the differences in the evaluation of parks in different zones, we divided the 50 parks into four categories: parks within the range of all green belts (18), parks in the first green isolation belt (8), parks in the second green isolation belt (15), and parks in the ring of forest wetland parks surrounding the capital (9), according to the relevant national policies. There is no significant difference in the quality of the urban parks in the four types of areas, but in terms of the composition of park categories, it can be seen from Table 1 that the parks within the range of all green belts and in the first green isolation belt are mainly comprehensive parks and humanistic parks, while the parks in the second green isolation belt and the ring of forest wetland parks surrounding the capital are mainly forest parks and wetland parks. The average score (number of stars) of the parks in each of the four regions was calculated separately for each year (Figure 15). According to the analysis, the overall average rating of all green belts, the parks in the first green isolation belt, the parks in the second green isolation belt, and the parks in the ring of forest wetland parks surrounding the capital in each period from 2006 to 2021 was 4.53 stars, 4.58 stars, 4.52 stars, and 4.43 stars, respectively. Note that the overall evaluation of the parks showed a sharp upward trend from 2008 to 2010, while in 2010, the ratings were basically stable, at 4 stars, and then, the rating steadily increased. The rating ranged from 4.45 stars to 4.75 stars in 2021. The reason for this phenomenon may be that the scoring function of the software platform was not yet popular in the early years, and the amount of scoring data was small.

Moreover, the star rating of the parks in all four zones decreased, but the score difference was not apparent.

The change chart of high-frequency words analysis of text data (Figure 16) showed that people paid more attention to keywords such as “environment” and “ticket” when evaluating parks within the range of all green belts, showing that visitors attached importance to the overall environment and service convenience of the park. The use of the words “Yuyuantan” and “cherry blossoms” has been increasing in recent years, indicating that the concentrated plant landscape in a certain period was a significant attraction to visitors. The frequent usage of the terms “Temple of Heaven,” “Beihai,” “The Forbidden City,” and “Jingshan” shows that travelers continued to pay attention to and placed value on these well-known picturesque locations.

People paid more attention to keywords such as “tickets,” “environment,” “child,” “forests,” and “running” in the evaluation of parks in the first green isolation belt (Figure 17), indicating that people paid high attention to exercise and parent–child activities in addition to the landscape and touring experience. “Forests” and “running” were hardly mentioned in the evaluations. This reflects the increasing demand of people in recent years to be close to the natural forest landscape and to execute sports activities in parks.

Evaluation of parks in the second green isolation belt (Figure 18) shows that people paid more attention to keywords such as “Fragrant Hill,” “red leaves,” and “mountain climbing,” which indicates that people paid a lot of attention to taking in the scenery and exercising. It also reflects that attractions with characteristic natural landscapes appeal more to the public. The attention to “parking lot” and “free” continued to rise every year, demonstrating people’s concern about the charges of parks, and visitors were more and more inclined to choose to drive to parks far away from their living places for activities.

In the evaluation of the parks in the ring of forest wetland parks surrounding the capital (Figure 19), people focused more on keywords such as “wetland,” “bike,” “wild duck,” and “child,” indicating that the parks’ wetland landscape, plant landscape, sports activities, parent–children activities, and other attractions were appealing to people. The attention to “parking lot” also continued to rise each year, demonstrating that the strong attraction of these forest wetland parks made visitors more and more inclined to choose to drive to parks far away from their residences. In recent years, “wild duck” and “forests” have drawn more attention, while “barbecue” has drawn less attention, suggesting that the biological landscapes of parks are now more valuable.

We counted the average number of reviews from January to December for the 50 parks in the four types of locations from 2006 to 2021 (Figure 20) and found that, except for October, the average number of comments for the parks within the range of all green belts ranks first every month, followed by the parks in the first green isolation belt and the second green isolation belt, and finally, the parks in the ring of forest wetland parks surrounding the capital. It is evident that the position of the green belts is connected to the variation in the typical number of comments for a park. The parks within the range of all green belts ushered in a peak in the number of comments in April. In addition, the parks in the second green isolation belt received the maximum number of feedbacks in October and November, and the parks in the ring of forest wetland parks surrounding the capital received the highest number of comments in May and October, while the number of reviews for the parks in the first green isolation belt was average.

It can be deduced that people are more likely to visit parks within the range of all green belts almost all year round, which might be linked to the fact that these parks are the closest to their homes. In October, people preferred parks in the second green isolation belt and parks in the ring of forest wetland parks surrounding the capital, which might be associated with the autumn landscape of these parks.

4.6. Differences in the Number of Comments on Parks at Different Times

To study the difference in the number of visitors to these parks on holidays and non-holidays, we counted the total holiday and non-holiday comments on all parks from 2006 to 2021 (Figure 21). The proportion of holidays and non-holidays in China was approximately 1:2. Analysis shows that all parks had more comments on non-holidays than holidays, which indicates that people have strong recreational needs even on weekdays and go to parks for their daily leisure activities on weekdays. The ratio of total comments on holidays and non-holidays for Yuyuantan Park, the Black Bamboo Park, Fragrant Hills Park, Taoranting Park, Beihai Park, Jingshan Park, Temple of Heaven Park, etc., was approximately 1:2. This suggested that rather than being exclusively vacation locations, these parks are now often used by people for daily activities.

4.7. Evaluation of Visitors’ Recreation and Sensory Perception Based on IPA

In an IPA (

Table 4. Importance and satisfaction scores of recreational indicators in different types of parks.

Indicators	Comprehensive Parks		Humanities Parks		Forest Parks		Wetland Parks
	Importance	Satisfaction	Importance	Satisfaction	Importance	Satisfaction	Importance	Satisfaction
A1	0.0834	0.6768	0.0764	0.7165	0.1169	0.7447	0.1373	0.6535
A2	0.1070	0.6662	0.0820	0.7121	0.0865	0.7411	0.0463	0.6488
A3	0.0378	0.7332	0.0334	0.7550	0.0464	0.7437	0.1356	0.7298
A4	0.1547	0.6436	0.1136	0.7244	0.1470	0.6893	0.1644	0.6246
A5	0.0361	0.6888	0.2000	0.7875	0.0130	0.7403	0.0077	0.6089
B1	0.0261	0.5927	0.0202	0.6375	0.0011	0.7922	0.0009	0.6296
B2	0.0545	0.6738	0.0294	0.6491	0.0744	0.7213	0.0563	0.6376
B3	0.0606	0.6389	0.0524	0.7384	0.0554	0.7603	0.0522	0.6798
B4	0.0509	0.6434	0.0266	0.6626	0.0732	0.7615	0.0491	0.6373
C1	0.0723	0.6687	0.0733	0.7371	0.0977	0.6915	0.0707	0.6453
C2	0.0597	0.6282	0.0257	0.6914	0.0928	0.6782	0.0714	0.5932
C3	0.0062	0.6289	0.0073	0.7238	0.0101	0.6659	0.0096	0.6122
C4	0.0195	0.6383	0.0113	0.7111	0.0078	0.7388	0.0154	0.6030
D1	0.1050	0.6123	0.0894	0.6709	0.0795	0.6956	0.0790	0.6087
D2	0.0738	0.6082	0.0950	0.7121	0.0541	0.6665	0.0711	0.6053
D3	0.0443	0.6182	0.0470	0.7404	0.0421	0.6870	0.0299	0.6089
D4	0.0083	0.6272	0.0043	0.7174	0.0019	0.7737	0.0031	0.6474

; Figure 22) of different types of parks, most factors of the comprehensive parks were in the third and fourth quadrants. This shows that comprehensive parks need to be improved in four aspects: landscape design, activity organization, infrastructure, and management. Among them, specifically, “mountain and water landscape” and “consumer spending” need to be improved, which may be related to the low-cost performance of comprehensive parks. Most factors of the humanities parks were in the first and second quadrants, with only “consumer spending” in the fourth quadrant. This shows that people were satisfied with all aspects of the humanities parks but wanted to control their consumption costs. Most factors of the forest parks were in the first and second quadrants. This result, together with the park evaluation score, proves that people were the most satisfied with the forest parks but that the park management still needs to be improved. The two factors of “animal landscape” and “recreational activities” were located in high-satisfaction areas for wetland parks, but these parks need to be improved in other aspects, especially in infrastructure. This may be related to the late construction of the wetland parks and the imperfect management.

In an IPA (

Table 5. Importance and satisfaction scores of recreational indicators in parks at different locations.

Indicators	Parks within the Range of All Green Belts		Parks in the First Green Isolation Belt		Parks in the Second Green Isolation Belt		Parks in the Ring of Forest Wetland Parks Surrounding the Capital
	Importance	Satisfaction	Importance	Satisfaction	Importance	Satisfaction	Importance	Satisfaction
A1	0.0745	0.7097	0.0940	0.7444	0.1200	0.6925	0.0866	0.6913
A2	0.1048	0.6832	0.0979	0.7643	0.0788	0.6265	0.0639	0.6635
A3	0.0371	0.7890	0.0398	0.7813	0.0497	0.6666	0.0545	0.7341
A4	0.1447	0.7006	0.1146	0.7411	0.1306	0.6278	0.1544	0.6409
A5	0.1626	0.7398	0.0201	0.7163	0.0402	0.7070	0.0321	0.6911
B1	0.0204	0.6148	0.0220	0.5065	0.0012	0.6967	0.0335	0.6628
B2	0.0259	0.6468	0.0670	0.7121	0.0841	0.6692	0.0489	0.6749
B3	0.0613	0.7179	0.0791	0.7359	0.0411	0.6822	0.0476	0.6805
B4	0.0316	0.6789	0.1150	0.7602	0.0451	0.6646	0.0402	0.6653
C1	0.0657	0.7489	0.0868	0.7703	0.0902	0.6285	0.0863	0.6356
C2	0.0249	0.7070	0.0604	0.7063	0.0922	0.6332	0.0779	0.5964
C3	0.0052	0.7396	0.0046	0.7838	0.0105	0.6000	0.0121	0.6269
C4	0.0135	0.7274	0.0104	0.7807	0.0113	0.6134	0.0227	0.5585
D1	0.0994	0.6570	0.0951	0.6467	0.0866	0.6519	0.0808	0.5949
D2	0.0807	0.7234	0.0439	0.6830	0.0744	0.6012	0.0914	0.6272
D3	0.0441	0.7220	0.0463	0.6018	0.0411	0.6574	0.0504	0.6208
D4	0.0035	0.7117	0.0030	0.6560	0.0030	0.7151	0.0166	0.6380

; Figure 23) of parks at different locations, most factors of the parks within the range of all green belts were in the first and second quadrants, but “activity organization” and “consumer spending” needed to be improved. Due to the large number of visitors, such parks may pay more attention to landscape maintenance rather than entertainment functions. In addition, landscape design, activity organization, and infrastructure of the parks in the first green isolation belt were in the first and second quadrants, while management aspects were mostly in low-satisfaction areas. This shows that the parks need to reasonably improve services to provide people with a more comfortable and convenient recreational experience. Most factors of the parks in the second green isolation belt were in the third and fourth quadrants, indicating that these parks still have a lot of room for improvement in all aspects. At the same time, because their area is generally large, it is particularly necessary to optimize the design of the guide service of the parks. Finally, for the parks in the ring of forest wetland parks surrounding the capital, in terms of landscape, most factors except “mountain and water landscape” were located in high-satisfaction areas. This may be related to the relatively high expectations of people from the landscape. Other factors were mostly in the third and fourth quadrants. People showed lower satisfaction with the factors of “transportation accessibility,” “public service facilities,” “food and beverage facilities,” and “consumer spending,”, etc., which may be related to the late completion and immature maintenance and management system of such parks.

According to the IPA (

Table 6. Importance and satisfaction scores of sensory indicators.

Senses Term	Indicators	Importance Degree	Satisfaction Degree
Vision (E)	E1	0.3390	0.7098
	E2	0.2267	0.7080
	E3	0.1435	0.7082
	E4	0.0863	0.7089
	E5	0.0185	0.6888
	E6	0.0052	0.7173
Hearing (F)	F1	0.0332	0.7013
	F2	0.0047	0.7074
	F3	0.0029	0.7172
	F4	0.0009	0.7042
Smell (G)	G1	0.0304	0.7055
	G2	0.0084	0.7142
Touch (H)	H1	0.0342	0.7121
	H2	0.0089	0.7137
	H3	0.0025	0.6926
	H4	0.0010	0.7077
	H5	0.0004	0.7019
Taste (L)	L1	0.0532	0.7087

, Figure 24) for different sensory perception, people’s demand for vision perception was significantly higher than that for other sensory perception, accounting for 81.91%; the demand for taste, touch, and hearing perception accounted for 5.32%, 4.70%, and 4.18%, respectively; and the demand for smell perception was relatively small, accounting for 3.89%.

In terms of vision, people were the least satisfied with the “vision of humans,” probably due to the large number of visitors, because of which people were unable to get a high-quality tourism experience. The taste factors only included “food available,” and people were more satisfied with it. In terms of touch, the “touch of roads” and “contact with animals” were in the third quadrant, indicating that managers need to focus on the rational planning and design of roads and animal protection measures in parks. In terms of hearing, people were disgusted with the “sound of voice.” This may be due to the heavy traffic and the disordered sound in the parks, leading to a poor experience. In terms of smell, people preferred the “smell of plants” but gave a poor evaluation to the “smell of air and water.” This may be related to the unstable air quality in Beijing. In general, managers should pay more attention to controlling the number of park visitors and improving park management, which will improve people’s sensory experience.

5. Discussion

This study was based on social media data collected from 50 well-known parks in Beijing, China. We compared park evaluations given by different genders and different geographical populations, park evaluations in terms of different attributes, and park evaluations across years and locations. Then, we summarized the public feedback on urban park construction and used big data analysis method to offer a foundation and point of reference for optimizing the functions of Beijing’s parks against a backdrop of urban regeneration.

In previous studies on park evaluation indicators, Calvin et al. [32] found that factors such as lawn, waterscape, wild animals and plants were more likely to attract people’s attention. Liu et al. [48] found that factors such as park scale, air quality, vegetation, mosquitoes, amusement facilities, sign systems, landscape visual effects, maintenance of facilities and plants, and environmental cleanliness had a significant impact on people’s satisfaction with urban parks. Combining previous studies on park evaluation indexes and our data analysis, we found that in terms of recreational elements offered by parks, people tended to focus on four aspects, landscape, activity, infrastructure, and management, with park landscape dominating the recreation experience, followed by infrastructure and management services. In terms of sensory perceptual elements, people tended to describe their personal leisure experience from several aspects, among which visual perception was the most important experience during people’s visit, while people generally did not pay attention to the auditory experience. Finally, we formed the evaluation system of park recreational elements with 4 major indicators and 17 factors and the evaluation system of sensory perceptual elements with 5 major indicators and 18 factors.

In terms of areas within the parks, the positive evaluations of Beijing parks primarily focused on the environment, scenic spots, and activity organization; the negative evaluations mostly focused on ticket supply, consumption expenditure, management services, etc. Recently, there has been increasing focus on parking lots, children’s activity spaces, and recreational facilities. This finding supports previous studies [48]. Thus, improving the infrastructure, richness, and variety of spaces is key to improving the citizens’ park visiting experience.

From the perspective of different groups of people, men placed greater emphasis on the convenience of park traffic, while women placed more emphasis on the consumption cost of parks and the adaptability of youngsters and the elderly. Thus, the needs of people of different age groups must be considered to create diverse recreational spaces. Comparing non-local visitors with local visitors, it was found that local visitors focus more on issues such as parking lots, while non-local visitors pay more attention to the characteristics of park attractions, ticket availability, and consumer prices. This is slightly different from the results of other scholars in other cities. Wang [45] et al. found that local visitors pay more attention to the landscape design of the parks than the infrastructure setting of the parks. This may be related to the difference between cities and the different quality of parks.

From the perspective of different park categories, people had different service needs from the four types of parks. Comprehensive parks must pay attention to accessibility by transportation as well as the overall activity area. Humanities parks must focus on fully demonstrating and expressing the history as well as culture by organizing science popularization cognition activities. The operation of parent–child activities should be the main emphasis of forest parks. Wetland parks must optimize ecological environment experience and strengthen ecological science education function. This is rarely discussed in the previous studies.

Our study also found something that previous research had never mentioned. From the perspective of different time periods, parks received more reviews for non-holidays than holidays. People had used parks as recreational places for their daily leisure activities. With passing years, people have started focusing more on parks in terms of the organization of traditional festival activities and the spending, and today, people look at transportation convenience, management, operation services, and so on. Thus, today, the construction of an urban park must be centered on the people and keep pace with the times to meet more diverse and rich needs.

From the perspective of different dimensions of time and space, people were more inclined to go to parks within the range of all green belts in the spring and to the parks in the second green isolation belt and the ring of forest wetland parks surrounding the capital in fall. The parks in the first green isolation belt had less fluctuation in visitor numbers throughout the year. Thus, when optimizing and upgrading urban parks in Beijing, we must fully consider the strong attraction of the seasonal characteristics of plants in spring and autumn for visitors and actively construct characteristic plant landscapes and seasonal landscapes in parks to achieve the efficient use of parks, to promote urban renewal, and to enhance people’s sense of happiness.

According to the importance–performance analysis of the evaluation index of recreation, regardless of the category and location of the parks, the factors that reduced the satisfaction of visitors were concentrated in the infrastructure and management, especially in the “consumer spending,” of the parks. According to the importance–performance analysis of the evaluation index of sensory perception, people were dissatisfied with the experience when there were a large number of visitors. This is similar to previous studies. For example, Liu [41] et al. proposed that the infrastructure construction of recreation space would affect visitors’ recreation satisfaction, but our study found more influencing factors. Zheng [49] et al. proposed that crowd disturbance would reduce the comfort of visitors’ experience of sensory perception during park recreation. Therefore, managers should actively promote the park infrastructure, improve the management service level, reasonably control the number of visitors to the park, and provide people with a more high-quality visitor experience.

The renewal and upgrading of urban parks play an important role in the renewal and optimization of urban habitat [50]. At the same time, the practice of public participation in the process of urban renewal is very effective and necessary [51,52]. The study used big data analysis to compare and analyze the evaluation contents of 50 Beijing parks on well-known social media platforms in various aspects, overcame the limitations of previous traditional park evaluation studies, and made suggestions for optimizing and improving park green spaces in the face of China’s urban renewal era.

In general, external factors such as accessibility and availability of tickets have been the main focus of people when visiting parks, which is largely in line with the findings of other scholars [25,32,48,53,54]. This study also found aspects not mentioned in previous studies: people now pay more attention to the organization of park activities, the arrangement of daily leisure facilities and the presentation of characteristic plant landscapes, and how to shape “park IP” and form “park brands” is the main direction for improving and optimizing urban parks. Zhang [55] et al. pointed out that park construction could promote the renewal of cities in both material and immaterial aspects. Similarly, our research found that by optimizing and improving various functions, urban parks continuously support and serve urban renewal decisions, improve the living environment, and enhance urban vitality, while coordinating with other urban planning projects to realize the comprehensive value and sustainable development of the city and let people enjoy more fruits of urban renewal.

6. Conclusions

This study selected Beijing, one of the first batches of urban renewal pilot cities in China, and used the big data text analysis method to analyze the scores of and comments on 50 popular Beijing parks selected by the public on social media. The significance of the study was that it avoided the shortcomings of traditional park evaluation studies and used new technologies to obtain a large amount of comprehensive evaluation information, making the results more timely and generalizable. Compared with many previous studies based on big data text analysis, this study broadened the scope of the study, circumvented the interference caused by external factors in individual park studies on park evaluation, and broadened the study time to discuss the impact of time change on park evaluation changes. Based on the dimensions of time and space, the study compared the evaluation results of urban parks from multiple perspectives, such as the characteristics of different people, various park types, the spatial and temporal distribution characteristics of recreational activities, forming a diversified research result. In addition, the study explored people’s comprehensive experience of the park by combining the IPA method and completed an in-depth investigation of the future direction of park optimization from two perspectives: recreation and perception.

However, many studies have pointed out that, while web platform data can provide access to more valuable information in a short period, they also have shortcomings [45,49,56]. The population involved was relatively simple because the majority of big data text information was provided by young and middle-aged people, with little input from children and the elderly. Thus, traditional offline questionnaires specifically targeting the elderly and children are also needed in the future to fill the gaps in data content for this population. A combination of online collection and offline surveys will help refine the study. Simultaneously, the semantic analysis network was only a display of some high-frequency influencing factors, and it was difficult to cover all factors. This needs to be avoided in the future by refining software development modules and making them more precise. In addition, there were some small inaccuracies in the updating of social media usage and operation techniques. In the future, the systematic bias caused by a single data source can be effectively reduced by means of multiple data crossover.

There are some shortcomings in the results of the current research, but with the development of information technology and the increasing number of Internet users, the differences between data obtained through social media and social survey data will gradually narrow, data analysis accuracy will gradually improve, and comparative research on urban park reviews based on big data will have higher operability. It is expected that subsequent studies will further improve the evaluation method of the park system and make more scientific and reasonable suggestions for the functional optimization of China’s urban parks, which will play a non-negligible role in China’s urban renewal actions.