What Role Do Urban Parks Play in Forming a Sense of Place? Lessons for Geodesign Using Social Media

Abstract

The sense of place is a multidimensional construct that evokes an emotional commitment to a specific geographic setting. It can be a crucial aspect of cultural ecosystem services. While social media has gained popularity as a tool for assessing ecosystem services, its effectiveness in capturing a sense of place, its impact on cultural ecosystem services, and its role in the landscape design process remains less certain. This study investigates the role of urban parks in shaping the sense of place by analyzing user-generated content from a specific social media platform (Twitter). We gathered tweets from 30 diverse urban parks in Chicago, covering various park types, sizes, shapes, and management styles. Our analysis reveals multiple facets of the sense of place associated with urban parks. We suggest that a sense of place is not solely rooted in the attachment to physical surroundings but also in the personal experiences individuals encounter within these spaces. Residents residing near parks tend to develop a sense of ownership and responsibility, leading to stronger emotional bonds with their environment. Urban parks foster community engagement, enhance social cohesion, and offer opportunities for nature-based experiences. Furthermore, this study underscores the significance of diverse park features, accessibility, and size in bolstering place attachment. Our research demonstrates the potential for geoinformation analysis in the geodesign process as a cost-effective and scalable approach for understanding the person–place connection.

1. Introduction

In an effort to promote low-carbon living, many of the United Nations Sustainable Development Goals (2030) target urban places. They aim to make cities more inclusive, safe, resilient, sustainable, and generally healthier in an effort to promote urban residential well-being [1]. To achieve these goals, they note that designers and planners should carefully balance the conflicts and the potential complementarity of the interests between the economy, environment, and social equity [2]. Geodesign-based landscape approaches have been found to be effective in integrating these factors [3,4] by providing an interdisciplinary platform for engaging both the material (through scientific analysis), and non-material resources inherent in the design process [5,6]. Among the non-material resources that might help promote residential well-being, developing a communal sense of place has sparked interest from a wide spectrum of researchers. Disciplinary research from ecology [7,8,9], human geography [10,11,12], environmental psychology [13,14,15,16], urban sociology [17], landscape [18,19,20], and urban planning [21,22] have published substantial work on the “person–place” bond.

In Geodesign processes, the people/place relationship has been noted as important for use in engaging differing perspectives [23], or understanding the spatial culture of a place [24]. Theoretically, the concept of a ‘sense of place’ is typically noted as a multidimensional construct that includes emotional ties and behavioral commitments to a specific geographic setting [11,13,14,25]. But because it cuts a wide disciplinary swath in the literature, multiple constructs and contradictory definitions have been proposed that can make this concept confusing, messy, and elusive [17,22]. For example, the sense of place might be explained as place attachment that can motivate community preservation [26], or environmental action [7]. Admittedly, the phenomenological nature of the concept contributes to its attractiveness as a way of illuminating a wide variety of experiences. Dissecting such a multidimensional phenomenon with too fine of a comb may result in loss of the “essence of the overall concept” [13].

Early instances in the literature describe a sense of place that emerges from “the perspective of experience” that emphasizes human emotions and relationships within a landscape [11]. Building on this, human geographers emphasize characteristics that are a part of a dynamic, temporal, and constantly changing lived experience [27]. Subsequent studies are more typically aimed at understanding a collection of meanings, satisfactions, and attachments across space and time—about how individuals or groups engage with the world around them [9,28]. Data for these studies are often collected through mail surveys to sampled residents at different spatial scales including towns [10,15], cities [29], and counties [30]. The response rate to these surveys is usually limited because only those that care deeply about the place typically respond [31]. This potential bias can result in a stronger sense of responsibility in measuring environment attitudes toward the place. Hannon (1994) also suggests that a sense of place denotes “expressions of desire” based on “geographical or spatial discounting” [7]. In this conception, people prefer to distance themselves from objects they fear and draw close to things that they desire. Hausmann (2016) suggests that this type of discounting can be viewed as a benefit to both biodiversity conservation and human well-being as it shows that living things are drawn to places that provide desirable benefits [8].

The most frequently cited definition of a sense of place is given by Stedman (2001). It is based on a combination of environmental psychology and social science concepts that consider a sense of place as an integration of place-based cognitions, identity, and attitude. It is defined as “a complex psychosocial structure that organizes self-referent cognitions, emotions, and behavioral commitments” that integrates affective, cognitive responses to the “space” or landscape [13,30]. Place attachment, place identity, and place dependence are considered sub-genres of the theoretical frame [16,31,32,33]. Generally, a sense of place is firmly associated with a sense of responsibility for a place [7,17,34]. Studies in conservation science also point out that the sense of place or place attachment is a part of a range of attitudes that encourages public engagement and civic responsibility [35,36].

Although the concept of sense of place varies by discipline, most conceptions appear to agree that a sense of place promotes a concern about a sustainable environment. Environmental psychologists believe that sense of place influences the public willingness to participate in maintaining or enhancing the valued attributes of the setting [16,30]. Alessa et al. (2008) mapped social–ecological space as a way of examining the sense of place [37]. They suggest that a sense of place is a crucial first step for maintaining community resilience, as it improves residents’ capacity for responding to disturbance. Brown et al. (2003) noted that place attachment increases the stability, familiarity, and security of the community, and can be adopted as a tool for neighborhood revitalization [38]. Most of the above definitions regard the sense of place as demonstrable, that it can be investigated with positivistic approaches including hypothesis testing and statistical methods. There are a number of studies that interpret a sense of place from a social–ecological perspective [15,18,29,34], tying the concept to notions of ecosystem services and cultural ecosystem services [9,22]. This has helped ground the ideas for use in broader design processes, including geodesign [4].

This study proposes a potential supplementary method for investigating the facets of a sense of place that emerge in urban settings—specifically, from urban parks on a neighborhood scale. This work utilizes city-wide, user-generated content generated from social media posts generated in urban parks within the Chicago, US, city limits. This work attempts to address two main research questions: (1) Is social media data effective in identifying a sense of place? And (2) what role do urban parks play in forming a sense of place in communities? The findings are expected to make statements on the feasibility of applying social media in investigating the person–place bond. They will also provide some implications for applying notions of a sense of place to geodesign and other design processes.

Following this introduction, we use the literature to review the concepts of a sense of place from a cultural ecosystem services, geoinformation, and urban green space perspective (Section 2). In Section 3, we describe our methodology, including the data collection, building of the interpretation framework, encoding analysis, and the model construction. An interpretation of the results within the three dimensions of the sense of place are illustrated in Section 4. In Section 5, we discuss the role of urban parks in promoting a sense of place, how this process is delivered via social media, some limitations, and some of the implications for geodesign. Finally (Section 6), we provide our concluding thoughts on the role of social media in geodesign along with the next steps for research.

2. Literature Review

In the introduction, we referred to some of the literature surrounding the conceptualization of a sense of place from a disciplinary perspective. In the following, we discuss it from a cultural ecosystem services viewpoint, followed by a brief review of the literature on geoinformation and its relation to social media.

2.1. Cultural Ecosystem Services and Urban Green Space

Urban green space is believed to play an important role in creating and fostering a sense of place [7,11,16,32], primarily due to its nature as a semi-natural space and its accessibility to a broad range of users [39]. These types of spaces have been noted to provide a diverse array of biotic and psychological goods and services often referred to as ecosystem services [40,41]. The typologies of ecosystem services have typically included notions of non-material, life-fulfilling functions [42] or cultural services [43,44]. The Millennium Ecosystem Assessment (MA) report (2005), for example, includes variables considered a part of ‘cultural ecosystem services’ including recreation, tourism, intellectual development, spiritual enrichment, and creative and aesthetic experiences—to name a few commonly cited [22,40,45,46]. The Assessment also includes a sense of place as a part of these considerations [40].

The Economics of Ecosystems and Biodiversity (TEEB) [47] suggests a framework for cultural ecosystem services that includes services termed “spiritual experience and sense of place”. It uses economic terms to describe a sense of place as a final demand-type service that is elicited from ecosystems [41]. Other literature connects the sense of place to cultural ecosystem services as more of an outcome. In this conception, ecosystem services provide the means for the emotional bonding process between individuals and their environment [7,25,26]. According to Wilson (1997), a sense of place emerges through active immersions with the environment, while the opportunities for engagement provided by the environment (services) determine the quality of these interactions [16]. Such opportunities include seclusion and quiet, active immersion with the natural world, personalizing spaces and places, and experience of memorable moments. Similarly, Scannell (2010) claims that urban natural space enhances the place bond with a restorative theme, which provides a secure and comfortable space for goal support, stress relief, and self-regulation [25]. From this perspective, the sense of place is not an independent cultural ecosystem service with a parallel position to other values like recreation or aesthetics, but is an outcome of experiences driven by multiple services.

Though the importance of nature in forming a sense of place has been frequently emphasized, how urban green space contributes to the sense of place building process remains somewhat vague and ambiguous. Additionally, assessing the person–place bond is context-specific, making it difficult to collect data at different scales. Traditional data collection methods like interviews, field observations, or on-site surveys are costly, time-consuming, and contribute to site-specificity [48,49,50].

2.2. Geoinformation, Social Media and Cultural Ecosystem Services

The literature widely acknowledges the potential of big data in enhancing our understanding of the built environment [51,52,53]. Advanced information technologies have enabled the investigation of places with a wide range of new data sources including data from sensors, monitoring, remote sensing, cell phone tracking, and social media [48]. These data, characterized by high volume, variety, and features like massiveness, rapidness, and complexity, present both opportunities and challenges in transforming it into valuable “information” [54].

Among the vast sources of big data, social media data hold distinct advantages in understanding person–place interactions with location identifications and easy access. Social media platforms serve as spaces where individuals share textual and visual content, often accompanied by geotags [55]. This wealth of information expands our comprehension of not only mobility patterns [56] but also individuals’ emotions, cognitions, and values towards the real environment over space and time [50,51]. Due to these characteristics, social media are recognized as a valuable resource for gaining insights into human behaviors, preferences, and experiences concerning the built environment. Consequently, social media data have been increasingly utilized in diverse scientific domains, including urban planning, marketing, tourism, ecology, conservation science, and disaster management [51]. Traditional data collection methods focused on subjective topics often involve direct observations, interviews, and visitor surveys. However, these methods are time-consuming, expensive, and limited in geographic coverage [49]. Therefore, exploring social media data to obtain valuable information about the sense of place within urban parks is a promising avenue for research.

Previous studies have presented compelling evidence that social media data, although representing only a fraction of the whole, can serve as a proxy for visits and mobility. Location-based social networking data, for instance, have proven useful in identifying urban travel flows [56] and estimating crowd sizes [57]. In recent years, multiple sources of social media have been employed to assess the accessibility and visit preferences related to urban green spaces [50,58,59,60,61,62]. Many of these studies have combined social media-based estimations with validations, demonstrating a positive association between the estimated metrics and visitor numbers derived from on-site observations, administrative surveys, and other indicators such as ticket sales [57,58,59,60].

Social media have been utilized to investigate various aspects of cultural ecosystem services. For example, aesthetic services have been studied with check-in data, user-generated texts, and social media images. Yoshimura and Hiura (2017) developed a method to map aesthetic value using geotagged photos on Flickr [63]. By considering topographical characteristics, they determined aesthetic demand scores based on factors such as the number of landscape photos, the number of photos taken at the same location, and the viewshed area from each photo [63]. Similarly, Langemeyer et al. (2018) proposed a spatial model for assessing aesthetic value by mapping the perceived landscape expression in geotagged photos shared on Flickr. This approach enabled the assessment of landscape aesthetic capacity and flow at a regional scale, a task nearly impossible with traditional surveys [64]. In terms of recreational value, Gosal et al. developed an approach utilizing Flickr data to identify recreational groups through image content analysis and latent semantic analysis [65]. Furthermore, studies in conservation science have suggested frameworks to assess the spatial distribution of multiple cultural ecosystem services using photographs shared on social media platforms in natural parks [62,66]. Social media have also shown insights in the understanding of landscape perception. Song et al. (2020) investigated how people use and perceive urban parks through self-posted, volunteer-created reviews on social media [67]. They developed a model using techniques like latent Dirichlet allocation to determine semantic topics emerging from reviews. In another study, Song et al. (2021) applied this approach to interpreting a sense of place with an environmental psychology lens [21].

3. Methodology

To examine the sense of place associated with urban parks, we collected user-generated content from Twitter (San Francisco, California, United States) posts that were generated within 30 parks in Chicago, IL. The parks represent the 30 most popular (according to number of relevant Tweets on Twitter) from an initial set of over 300 urban parks in the city. They cover a diversity of park type, size, shape, and management—from small neighborhood parks and community parks, to big regional or citywide parks and represent a good sampling of Chicago urban parks (see Appendix A).

Our usage of crowdsourced social media data comes with some limitations: the representativeness of this study might be confined to economically advantaged, well-educated, urban-dwelling young adults as Twitter users. One critical component of using social media in a specific location is the use of smart phone technology. According to the Pew Research Center (2021), the majority of US adults (85%) owned a smartphone in 2021, with 72% of the public using some form of social media [68]. This survey also reports that 42% of internet users between 18 and 29 have ever used Twitter. Pew also notes that 30% of internet users have used Twitter, but this number rapidly drops to 12% for those who earn less than 30 K per year. As of 2016, there are over 330 million monthly active users on Twitter [69]. These numbers have been increasing over the past decade and continue to rise steadily [53]. Also, crowdsourcing geoinformation on social media is biased towards urban perspectives [70,71], while they offer generalizable and comparable information in densely populated metropolitan areas [69,72]. This gives us some confidence that our study will represent a reasonably large sample of park users, although its overall generalizability is questionable.

Data were collected through the Twitter API for approximately 1 year, from March 2021 to April 2022. Park names were used as search keywords across the entire city of Chicago to collect relevant tweets (approximately 22,600 were collected). Recognizing the importance of capturing shared experiences after leaving a park, we also included tweets that described park experiences not only from individuals present within the park or its immediate vicinity but also from those including references to the park. For each record, we collected the user ID, status ID (unique to each tweet), creation time, and text content. Among the 30 selected parks, Lincoln Park emerged as the most popular, while Kelly Park ranked lowest in terms of mentions (Figure 1). The number of tweets posted showed a seasonal trend, with a higher volume during the summer months. However, there was no significant difference between weekdays and weekends (Figure 2).

Given the wide variety of content posted on social media platforms, it was necessary to establish specific criteria for selecting tweets that contributed to the person–place bond at a neighborhood scale, considering the extensive dataset. We employed 3 rules for filtering the tweets (see Appendix B 

Table A2. Examples of tweets selection.

Tweet Text	Action	Reason
I love Chicago in the fall 🍁. #chicago #lincolnparkchicago #compass #fall #autumn #door #home #neighborhood #leaves #thanksgiving @ Lincoln Park	Retain	The tweet is geotagged with the location inside Lincoln Park
A futute lil flower guy floating in the sky #Spring🌸🌸 @ Humboldt Park, Chicago	Retain	The tweet is geotagged with the location inside Humboldt Park
Walking in the neighborhood today we got to see all kinds of animals.#citylife #ourneighborsareanimals #lincolnpark #chiccago @ Alfred Caldwell Lily Pool	Retain	The tweet mentioned Lincoln Park, and the text indicates that the user spent time in the Lincoln Park neighborhood
I live in East Lincoln Park for the above reasons. Also, free zoo! Who doesn’t like that? Tons of greenery and parks in the neighborhood. Also, Absolutely ZERO shortage of coffee ☕️ shops 😂	Retain	The tweet mentioned the user’s preference living in the Lincoln Park neighborhood
Marquette is in a waaay worse neighborhood than Lincoln Park.	Retain	The tweet is geotagged with the location in Marquette Park
Your finest Chitown planners and your favorite community engagement associate 😜 we were out in Washington Park for a community event today! @cmapillinois @lindsaybanks #womeninplanning #communityengagement	Remove	Although the tweet is posted with a geo-tag inside Washington Park, it is published by an organization rather than a visitor
Summer cultural events across Chicago have been reimagined as part of #OpenChicago. New programs include “Chicago Presents,” “House City,” “Taste of Chicago To-Go,” Latinx; World Music community events—plus much more in Millennium Park.	Remove	Although the tweet mentions lots of experiences in Millennium Park, it is not edited with a first-person narrative, and we cannot decide if the user spent time on these activities. In fact, it looks more like an advertisement.
“The Southeast Side deserves to be known for its GREEN SPACES. No other neighborhood has as many parks, preserves, or natural areas like we do. And yes that is a flex 🌳 If you’re reading this and are able, go visit Eggers or Big Marsh and just listen to the birds talking”	Remove	The main text is quoting an article

for example tweets):

• The tweet was geotagged with coordinates located within the park or directly adjacent, indicating that the user was in or around the park at the time of the post. Consequently, retweets, quotes, and advertisements were excluded, even if geotagged.
• If a tweet was not geotagged, it should mention the park. The text should reflect the visitor’s experience inside or around the urban park or its neighborhood.
• The text had to be written in the first-person narrative and not posted by organizations, businesses, or authorities.

3.1. Sense of Place Interpretation Framework

The interpretation framework for the sense of place in our study is based on a broad concept that captures the meaningful relationship between individuals and their environments. This concept draws from various disciplines and is influenced by definitions and structures of sense of place, place attachment, and community attachment [11,25,30].

The foundation of our framework is derived from a tripartite organization proposed by Scannell (2010). It views the person–place bond as a multidimensional concept with person, psychological process, and place dimensions (Figure 3). The “Person” dimension focuses on the individuals or groups who are attached to a place. At the individual level, personal experiences, growth, and significant milestones contribute to the formation of meaning and evoke personal memories associated with a place. When these experiences are shared among members of a group, they further shape the symbolic meanings of the place [25]. Cultural place meanings and values also influence the degree of individual place attachment.

The “Place” dimension encompasses both the physical settings and the social context. According to Lewicka (2010), neighborhood bonds are formed through interpersonal interactions, while certain physical features and the social context can influence these interactions [14]. This dimension consists of three facets: “Built environment”, “Nature”, and “Territoriality”. The first two facets pertain to the physical settings, including the artificial constructions and natural elements within urban areas, which are central to the attachment process. This process is often referred to as place dependence, where the place provides amenities or resources that support individuals’ goals [13]. The “Nature” facet emphasizes the environment itself and the potential opportunities it offers, focusing on active immersion, exploration, and interaction with the natural environment. Additionally, the inclusion of nature in one’s self-concept contributes to the formation of an environmental identity [73]. The third facet, “Territoriality”, relates to the social arena and a sense of responsibility of a place and encourages the engagement and protection of the environment [7].

The “Process” dimension of the sense of place indicates how individuals and groups relate to a place and the nature of their psychological interaction. This dimension includes facets such as “Affect”, “Cognition”, and “Safety”. The “Affect” facet represents the emotional connection individuals and groups have with a particular place. The self-emotion topics encompass emotional connections such as love, happiness, joy, pride, as well as fear, hatred, or ambivalence. The “Cognition” facet refers to the memories, beliefs, meanings, and personalities associated with specific settings, contributing to a sense of closeness to a place [25]. Under this dimension, we identify three facets that are relevant to online interactions: “Memories”, “Certification”, and “Presentation”. “Memories” suggest a bond with a place that spans over time and has connections to the past. The other two facets involve personalizing space and places, “certifying” the importance of a place, and subsequently fostering a sense of “ownership” of that place. “Certification” refers to actions that highlight a particular place, increasing the awareness of its significance. The “Presentation” of a place involves activities such as videotaping, photographing, or descriptive writing, emphasizing the importance of that particular place [16].

3.2. Data Encoding Analysis and Model Construction

In our analysis, we used a process shaped by the views of a large number of participants in order to test the role of cultural ecosystem services in forming place attachment. We encode our crowd-sourced data in three phases using Nvivo 12. We first sampled 15% of the whole dataset to train the coding pattern based on our interpretation framework. The open coding phase begins with acquiring the direct element of place-related information from Twitter texts. Sample tweets related to Chicago parks and communities are imported into Nvivo 12 for text analysis, and they are coded into 319 free nodes. These nodes are located at the bottom of the affiliation, which indicates the direct information that people experience for the place. One tweet may be coded with two or more nodes, and the text contents are coded as much as possible. Then the axial coding and selective phrase interconnect the categories and develop the model. The free nodes are summarized and formed into 30 s-level nodes, and these second-level nodes are further generated into 8 first-level nodes as identified in the interpretation framework. These first-level nodes, combined and compared with theories in psychology and human geography, are then used to conduct the auto-coding process to the whole dataset.

To measure and compare the sense of place implicated in tweets, sense-of-place salience for each park is computed based on the number of tweets coded in the proposed interpretation framework. The primary assumption is that if people make a post that meets the four criteria listed in the previous section, they associate a sense of place with the park or the surrounding community. A park with more coded tweets in one topic implies that more people associate this facet of sense of place. We quantify and characterize the sense of place that is implicated on social media for each park based on the normalized score in each facet of sense of place, and then apply salience scores to build binomial regression models to examine whether certain facts of sense of place are more noteworthy in specific park types.

4. Results

Different facets of the sense of place emerge from the text data and they indicate the fundamental elements in forming a sense of place. In total, 763 out of 22,615 tweets are identified with nodes relating to the sense of place interpretation framework. Figure 4 shows tweets coded to the interpretation framework by park type. For each bar plot, the y-axis lists the park type, and the x-axis shows the number of coded references in this given dimension. The difference is significant among park size: a magnet park may suggest far more power in creating a sense of place in all three dimensions; only a minority of the coded reference is from neighborhood parks or passive parks, while parks with more influence contain more user-generated contents that are related to the sense of place interpretation.

Table 1. Regression results for citywide parks, regional parks, and community parks.

	Citywide		Regional		Community
	Estimate	p Value	Estimate	p Value	Estimate	p Value
Cultural or group	0.33	0.98	7.56	0.61	−94.76	0.21
Individuals	60.65	0.10 *	−6.65	0.84	−21.26	0.45
Built environment	−759.71	0.09 *	−15.73	0.48	23.38	0.43
Nature	4.06	0.71	56.81	0.17	−32.26	0.05 *
Territoriality	52.41	0.20	−3.67	0.85	−26.95	0.14
Affect	−90.65	0.09 *	−61.43	0.21	48.83	0.17
Cognition	143.22	0.17	18.00	0.40	41.31	0.10 *
Safety	−13.22	0.24	−12.70	0.24	11.02	0.14

* = The significant results with p ≤ 0.1.

shows the regression results for citywide parks, regional parks, and community parks. A positive estimation value means they are positively associated, and the p value indicates the level of significance. The significant results (p ≤ 0.1) are highlighted in bold texts. Figure 5 visualizes the analysis with bar plots, where the x-axis lists the park types, and the y-axis refers to the estimate coefficients. The color of the bar represents the p-value: the darker the bar is, the more significant the association. The estimate coefficient of the “Built environment” in the citywide parks and regional parks is negative while positive in the community parks, which indicates that people care less about the artificial environment in larger parks than in smaller parks. Notably, all the facets in the “Process” dimension, including affection, cognition, and safety, show as positive factors in the community parks. Due to the limited sample size of the urban parks, the regression results for the magnet parks, neighborhood parks, and passive parks are not significant.

4.1. Person Dimension

The “Person” facet indicates the personal connections one has to a place, and this expression of symbolic meanings can occur at both the individual and group levels. The facet “Individuals” includes text related to activities, milestones, and realizations that contribute to personal growth or self-regulation (See Appendix B 

Table A3. Examples of tweets showing a sense of place in the “Person” dimension.

Facet	Semantic Theme	Tweet Text
Cultural or group	Demographic	Im a Black man from Garfield Park Chicago. Look that neighborhood up on Google real quick. Actually know what you are talking about before saying any ole shit about someone.
		I remember Wicker Park when 90% of those that live there now wouldn’t drive past there. Lincoln Park was a Puerto Rican neighborhood way back. Humboldt Park, Logan Square, it’s all gentrifying; while some people do respect older residents you can’t ignore the hostility we feel.
	Historical	My Neighborhood Riverwalk guide explains some historical fact of Chicago River on best late birthday celebration ever! @ Horner Park, Chicago
	Religious	Then, on a long afternoon bike ride on the North Branch Trail, the kids and I passed a large outdoor gathering at forest preserve. It took me a few minutes, but suddenly I started hearing Kyrgyz music, and then came to discover it was an Kyrgyz community Eid celebration!
	Other public events	A live performance! It’s been a long time. Even though the event was outside it was very welcome. Many thanks to the Grant Park Symphony. I love it.
Individuals	Activities	Went for a walk this morning to McKinley Park in my neighborhood and I was reminded of what I and so many other people (kids and families) have to breathe everyday if they want to go outside. It was really bad. #stopmatasphalt @aldcardenas
		Biking in #Chicago from Millennium Park to Lincoln Park neighbourhood (5 mile ride).
	Milestones	31derful years! Happy anniversary to my fair lady, Cheryl Sabas. Thank you, my love. I love you. @ Lincoln Park
		I’ve been chasing this race for years. Not only did I kill thr time, I finally made it. #runchi #running #shamrockshuffle @ Grant Park
	Realizations	A moment of rest and reflection at our community garden in Gage Park.

for tweet examples). Most of the personal activities are recreational or connected to a restoration theme. It is an important theme in forming a sense of place because people develop a person–place bond in a secure, comfortable environment as it is conducive to self-reflection, problem-solving, and stress relief [25]. Figure 6 shows these semantic themes by park type. “Activities” refers to a variety of outdoor activities including walking, biking, skating, football, basketball, etc., among which walking ranks first in frequency and is followed by biking. This theme is not coded in neighborhood parks or passive parks. Milestones and realizations are important personal experiences where people create meaning for the place. The facet “Cultural or group” emphasizes a “community process”: groups affirm “distinctiveness” based on class, race, and religion, and they become attached to a place where they practice and preserve their cultures. The demographic and religious themes tells of the local sentiment and ties of the “actor”, including age, race, income, family, and religious activities, which contribute to the place-based distinctiveness from other groups. Demographic and religious topics are frequently detected in citywide parks, while daily experiences like activities and realizations do not show significant differences among the parks. This status shapes the place identity and facilitates the specialness of one’s group. It agrees with the acknowledgment that place attachment is usually religious-based and has strong cultural fealty [25]. The remaining two themes, namely “historical” and “public events”, emphasize the shared experience of group members from the perspective of the past and present, respectively. These themes are less likely to occur in neighborhood parks and passive parks.

4.2. Place Dimension

The “Place” dimension contains both physical settings and the social arena. In our study, there are four facets claimed under this dimension that we can find references of on social media: “Built environment”, “Nature”, and “Territoriality” (See Appendix B 

Table A4. Examples of tweets showing a sense of place in the “Place” dimension.

Facet	Semantic Theme	Tweet Text
Built environment	Commercials	I live in East Lincoln Park for the above reasons. Also, free zoo! Who doesn’t like that? Tons of greenery and parks in the neighborhood. Also, Absolutely ZERO shortage of coffee ☕️ shops 😂
	Facilities and amenities	I like my new apartment and being close to transit but holy shit wicker park is least friendly neighborhood I’ve lived in.
	Street views	“Night Lights”. I loved the quality of the light hitting the houses in my Rogers Park neighborhood. @ Rogers Park, Chicago
Nature	Plants	There are some PRETTY trees in my neighborhood. @ Portage Park
	Waterbody	I drove from my neighborhood on the western end of Chicago all the way to the lake last Sunday and saw few outdoor masks in heavily populated areas like Humboldt Park and Bucktown but all young people. At least no one accosts me for masking!
	Animals	I found a bird in a tree yesterday. And we saw the first butterfly of the season. #outandabout @ Humboldt Park, Chicago
	Season	I love Chicago in the fall 🍁. #chicago #lincolnparkchicago #compass #fall #autumn #door #home #neighborhood #leaves #thanksgiving @ Lincoln Park
	Weather	Washington Park makes the snow look like a movie scene 😍
Territoriality	Community engagements	My little part of keeping the neighborhood relatively affordable and diverse.
	Concerns of the community	they don’t have a budget to invest in the Gage Park Library. They spent $257,760 renting a rundown space from a private landlord over the last 6 years. They have been renting for 30… bad financial planning. Buy a building in our community.
	Dwelling description
	Compare between neighborhoods	Marquette is in a waaay worse neighborhood than Lincoln Park.
	Neighborhood quality	@swingkennedy Oak Park (although it’s not in-Chicago, it’s still accessible via the green line or Metra). La Grange is also a fantastic suburb. As far as neighborhoods go, Rogers Park is a great family-friendly neighborhood. But Lincoln Park is the go-to in the city.
	Travel time or distance	@IrvLeavitt having owned 14 different homes in various neighborhoods over my life (grew up in Sauganash and despised it) my favorite was Edison Park...restaurants are plentiful...xpwy minutes away...cops and fire are top-notch neighbors...very little crime...

for tweet example). The physical environment is expressed in two facets, namely “Built Environment” and “Nature”, while the “Territoriality” emphasizes the virtual arena that is based on the ownership, control of space, and the regulation of access to self [25]. There are three themes clarified under the facet “Built environment”: facilities, commercials, and street views (see Appendix B Table A4 for tweet examples). Figure 7a shows the distribution of nodes in this facet. The topic “facilities” refers to the online discussions of urban facilities, and it covers more than half of the references coded within this facet. Based on our dataset, the top-mentioned facilities are plazas, transportation, libraries, and zoos. As for commercial places, restaurants and cafes occur most frequently. All these places are open to the public, and consequently, support social interactions and foster social ties rooted in place. Another theme related to the artificial environment points to individuals’ experiences and attitudes towards the street views, including the alleys, building facets, night lights, etc.

Nature is another significant part of the physical setting. Under this theme, we found that people express their appreciation for the community with the exact nature elements within or around urban parks, such as lakefronts, beaches, the Chicago River, blooms, and trees. The proximity to a waterbody is popular and sometimes claimed as a reason to stay. Meanwhile, people also love to share their detections and feelings about the change in season and weather on social media. These findings correspond with Wilson’s suggestions on developing a sense of place for child development: direct exposure to a variety of natural encounters encourages people to feel, hear, taste, smell, and see the natural world, which tends to foster a rich sense of place experience. However, little information about the soundscape is identified in our dataset and we failed to generate a theme for the acoustic environment. There are more tweets about natural elements found in regional parks, while none in neighborhood parks and passive parks (Figure 7b).

“Territoriality” describes the social arena. It is associated with a sense of responsibility of a place and encourages the engagement and protection of the environment [7]. People show their engagements, concerns, attitudes, hopes, and responsibilities on social media. “Concerns of community” covers concrete discussions about existing problems, proposals, or policies about education, financial plans, social justice, poverty, transportation, etc. for the community. In the theme “Community Engagement”, there are lots of records for community volunteer works, and also calls for engagement from local organizations. Residents also love to write down when they detect changes in the neighborhood as reconstruction vitalizes the neighborhood and reinforces a sense of place. “Neighborhood quality” is close to the concept of “place satisfaction”. Stedman (2002) states that a sense of place includes cognitions, identity, attitude, and potential behaviors, while place satisfaction represents the attitude towards a place [30]. It refers to the summary judgment of the perceived quality of a setting and predictive, and is associated with life quality [74,75]. As we obtained a relatively high portion of such tweets in our dataset, this indicator can be developed as a potential tool for designers with easy access. Urban parks are also used as a spatial reference of distance or travel time, which indicates that attachment is expressed through proximity-maintaining behaviors. In addition, people present urban parks as a location reference when describing things around them, especially for dwelling statements. Dwelling statements refer to the tweets where people claim themselves to “move into”, “move out of”, or “live in” a park, even though the administrative name of the community does not correspond with the park name. These kinds of statements are more likely to be found around those middle-sized parks like community parks, citywide parks, and regional parks. Residents use the urban park to compare their living community with others. This process again stresses the information about one’s distinctiveness or similarity provided by the place. These features make residents feel unique about their living place and attached to one’s self-concept. In other words, urban parks help residents to represent who they are. This semantic pattern occurs with a significantly higher frequency in community parks (Figure 7c).

4.3. Process Dimension

The “Process” dimension covers facets concerning the psychological interactions and functions of a sense of place (See Appendix B 

Table A5. Examples of tweets showing a sense of place in the “Process” dimension.

Facet	Semantic Theme	Tweet Text
Affect	\	I made it through Gage Park, Southside of Chicago!! I’m shedding light on my neighborhood and the circumstances you’re faced with as a Young Black Man 🤝🤧💔. This is where I’m from and I take pride in how far I’ve come 🌟🎯💯. Thank you @BeingMaryJhay 💗💗💗. You the realest⛽️💕.
Cognition	Memories	I grew up playing small fry basketball at Marquette Park Coach Green was somebody who taught me a little about myself growing up, today he passed away and I’m just thinking about home and the community that helped shaped me into the person I am today
	Certification	@KatContii If you come to the city. Logan square and wicker park neighborhoods will be your go to for food and beer
	Presentation	I took this photo during summer of this year. It was in a small restaurant in Wicker Park. If you know the area, you might know what it is. 😉
Safety	Security	I know some of them live in lower crime north side neighborhoods and suburbs. Logan Square, Wicker Park, and Avondale are all represented.
	Health	As more and more factories open in Gage Park I know that the health of our community is under attack.
	Other	Help me make my neighborhood safer.

for tweet example). The “Affection” facet represents individuals’ and groups’ emotional connection to a particular place. The self-emotion topic refers to emotional connections including love, happiness, joy, pride, and also fear, hatred, or ambivalence. Most of the self-emotion tweets are with positive descriptions. Our results agree with previous findings: usually, the self-emotions associated with place attachment are in positive terms though negative valence bonds form sometimes [25]. As the pattern-based coding method in NVivo is not good at detecting positive or negative affections, we do not divide themes under this facet. As shown in Figure 8a, affection expressions are delivered more frequently in community parks, followed by magnet, regional, and citywide parks, while few emotional expressions are in the neighborhood or passive parks.

As for the “Cognition” facet, the theme “Memories” suggests an experience that a certain setting contains in the past, and enhances the sense of place by facilitating closeness to this meaningful setting. In our study, the texts cover topics mainly about childhood experience and the change in the residential environment, which happens more frequently in regional parks (Figure 8b). The online content in the “Certification” theme focuses on the introduction and summaries of a park or neighborhood from a personal view. Landscape photos and video posts are the most common content in the theme “Presentation”. This topic is frequently coded in community parks (Figure 8b).

The facet “Safety” mainly contains concerns about security and health issues, and the topics are more likely to be found in citywide parks (Figure 8c). Contents coded under the security category are related to gun violence, crimes, robbery, and other public security concerns. The pandemic concerns consist of the largest portion of the health issues in our results. The “other” includes some general descriptions or hopes for the safety concept.

5. Discussion

We have entered a new data-rich environment where designers have access to more comprehensive, contextual, and complex data sets than ever before [48]. Social media in particular produce multiplicities of digital traces whose contents range from tweets to images, text, tags, and shared locations. These massive sources are now being used to gain insights into how we understand the city and human behavior patterns. An increasing number of publications have identified social media data with great potential for investigating when, where, who, and how people use urban open spaces [50,53,59,61,76]. However, this readily available and rich data stream could also be challenging due to its nature of also presenting some risks (see Section 5.3 below for a discussion on these challenges). This section discusses the ability of social media in revealing the people–place bond, implications for geodesign, and how urban parks contribute to forming a sense of place.

5.1. Social Media Data, Geodesign, and a Sense of Place

In our study, we use a machine learning approach of utilizing a small (training) dataset to code and apply (teach) our sense of place interpretative framework to the larger dataset. We start with big data, but our results show that only a small selection of the social media posts we gathered were useful for our sense of place interpretation. Posts that were successfully coded with facets of the sense of place accounted for only 1.73% of the whole (about 400). Although far less substantial, this subset was still useful in understanding the elements of the place bond.

The sense of place is not directly developed through attachment to the physical settings of a place; rather, it is created through processes of “experience in place” [25,30]. Cognitive processes, such as memories and significant moments, associate individual experiences with specific settings, making the physical features meaningful. These cognitions distinguish one place from another and facilitate a person’s emotional connection to a place. People often document memorable experiences on social media platforms, which contributes to the unique strength of using social media for sense-of-place analysis compared to traditional surveys. Additionally, the process of sharing descriptive writing or photography on social media reinforces the sense of place. The act of presenting and certifying experiences on social media increases the awareness and importance of a particular place [16]. As social media platforms provide functions for sharing text, images, and videos, they serve as great platforms for both individual and community processes related to the sense of place.

Social media platforms also provide a wealth of information about territoriality. People express their awareness of neighborhood changes, nostalgia for the past, and hopes for the future. Engagement and responsibility are also reflected through social media platforms, where discussions on existing problems, community events, and volunteer work take place. Strong feelings toward a place often prompt individuals to express their attitudes, and social media provides accessible platforms for communities to propose actions and call for engagement from residents. Given its easy accessibility and rapid response nature, social media can be utilized as a design tool for negotiating and addressing the needs of stakeholders, particularly those who have limited decision-making power. In geodesign, social media can be developed as a supplementary tool to quickly gather preliminary knowledge about a sense of place, with place satisfaction serving as an easily accessible indicator.

In this study, we find that social media can usefully inform the design process by providing a wide variety of data—in the case of geodesign, these data are useful for a range of representation and change models.

There are three commonly occurring problems in most landscape design processes: a lack of real-time assessment, a lack of dynamic analysis, and a lack of public participation [5]. Geodesign attempts to address these shortcomings with a series of defined modeling steps including representation models, process models, and evaluation models [77]. Since social media metadata records both geographical and temporal data points, designers can use social media to spatially describe place bonds in the representation models, and generate real-time simulations in the process and evaluation models. As for public participation, our findings demonstrate that although only a small subset of tweets claim a sense of place, it does provide a quick glance at the people–place relationship in metropolitan areas. Compared to other participatory processes, social media can generate large numbers of observations, record engagement and retention over time, and contain less selection bias on the users’ interests (e.g., surveys are more likely to receive responses from those who are interested in nature or landscapes) [78].

5.2. How Urban Parks Contribute to a Sense of Place

Urban parks play a crucial role in fostering a connection between people and a place. From a territorial perspective, residents living near a park often develop a sense of ownership or regulated access, which contributes to a sense of responsibility for the community. The name of an urban park is frequently used to represent the surrounding area where one resides. It is a common phenomenon for residents to refer to their community by the name of a nearby urban park rather than its administrative name. This tendency arises when individuals feel a sense of ownership over that place and may be influenced by the park’s size. Parks also provide information about a person’s distinctiveness or similarity. People compare different communities based on urban parks, even though the park’s name may not be the administrative name of the community. This language pattern is most frequently observed in community parks, followed by regional parks or citywide parks, and is rarely found in small neighborhood parks. In conclusion, residents are more likely to use community parks to describe their living status, thereby developing stronger emotional ties to the place.

Urban parks also contribute to local place bonding by providing citizens with access to the natural environment. On social media, we found people associate active nature-based immersion with parks, develop a sense of seclusion and restoration, and show responsible land stewardship as well. In addition, much volunteer work and many community hearings gather at urban parks in Chicago. Urban parks encourage community engagement and enhance social cohesion by providing opportunities for preservation practice.

Among the various parks studied, we found that parks with different administrative influences exhibit distinguishing features related to the sense of responsibility for administrative or demographic issues, such as community proposals, neighborhood satisfaction, facilities, and security events. Larger parks, associated with a greater historical influence, show higher levels of overall salience across various facets of sense of place. Notably, regional parks show a relatively positive association with expressions of the sense of nature (p = 0.17). Regional parks may strike a balance between nature opportunities and accessibility, as small urban parks fail to provide an immersive natural environment for adults, while large parks may not be as accessible for daily entertainment. Community parks exhibit more potential in connections related to personal psychological processes and residents’ daily routines, such as recreation, commercial attractions, descriptions of dwelling, concerns about the community, and neighborhood quality. In addition, small parks tend to have negative coefficients in terms of the sense of nature. For community parks and neighborhood parks, it is individuals’ feelings and bonds associated with exposure to nature that are noteworthy, rather than the natural environment itself. Passive parks lack open spaces that provide visitors with opportunities to create their own “experience in place”.

These findings provide designers and policymakers with new evidence to enhance place attachment in communities. Community parks ranging from 5 to 10 acres in size are more likely to facilitate affective ties within neighborhoods and contribute to social cohesion in residents’ daily lives. Parks that excel in multiple facets of the sense of place are characterized by ample space for immersive nature experiences, open areas for social or recreational activities, and accessibility for residents.

5.3. Limitations

Admittedly, findings drawn from social media may not be representative of all groups of people. First, communicative media data do not accurately reflect mobility. Users selectively share content when they post, which can be performative in nature. Furthermore, social media data often exhibit demographic biases based on user profiles [53]. Since we only conducted the text analysis on Twitter, there is a bias towards that particular platform, as users share different content on different platforms. The majority of Twitter users are well-educated adults with an annual income higher than $30,000 [68], thus limiting the representativeness of the findings to younger and older generations. Many sense of place studies have emphasized the role of local knowledge, engagement, and exposure to the natural environment during one’s childhood in developing a sense of place and self-identity. Due to demographic bias, we found little support for these ideas in the social media data used in our study.

However, we cannot ignore the powerful potential of social media as a tool for gaining insights into the perceived environment. Evidence suggests that visitation estimates generated from social media platforms, while not equivalent to actual visitation rates, can serve as reliable proxies for empirical visitation rates validated through on-site observations [50,58,60,61]. Given the unique advantages of social media data in terms of scale and near-real-time observations, there is an opportunity to investigate various aspects of place attachment across a wide range of urban green spaces. We argue that the values and bonds associated with place suggested by the sense of place literature can be detected in user-generated content on social media, making it a valuable supplementary tool for gaining quick insights into different facets of the sense of place.

In this study, we sampled approximately 15% of the dataset for manual coding and training. As a result, some themes in our predefined framework based on the literature were not generated, such as soundscapes and smells. Additionally, the Nvivo 12 software we used does not provide validation or detection of false patterns. Moreover, our approach failed to identify the place bonds shown in photo posts, which account for a significant portion of the data sources.

6. Conclusions

In this paper, we proposed a framework for detecting a sense of place using social media and demonstrated its potential in informing geodesign. We conducted a city-wide text analysis of urban parks to determine whether online user-generated content can be applied to identify a sense of place. We found that urban parks can play a pivotal role in fostering a sense of place by creating a sense of ownership, responsibility, and community engagement. They offer access to nature, contributing to immersion, restoration, and responsible land stewardship. We also found that park size and administrative use can influence their role in community life.

Our research demonstrates that social media data offer a valuable resource for geodesign by providing insights into the people–place bond and territoriality in urban areas. It compliments traditional methods and can address common landscape practice challenges, including real-time assessment, dynamic analysis, and public participation. It can serve as a supplementary tool for landscape designers to perform quantification, visualization, and prediction of public values in geodesign. However, it is essential to acknowledge the limitations of social media data, including potential biases and data selection challenges. Our findings are specific to young adults in metropolitan areas and may not apply universally.

Future research could focus on adopting new text analysis technologies, such as machine learning algorithms, to address the computational challenges posed by massive data streams. It would also be valuable to investigate the identification of a sense of place using shared photos on different social media platforms and examine how this approach may vary across different cultural contexts.