PDF Download
PubReader
Export Citation
Email the Author
Share Facebook
Share Twitter
Cited by 0 Article
Metrics (View:419)
1. Introduction
The term “New Towns”, as ubiquitously employed in the Korean context, denotes a constructed urban center at the periphery of the sprawling in the Seoul Metropolitan Area (SMA). New towns are designed to strategically alleviate population congestion within the highly populated area. The development of New Towns was initiated during the late 1980s and had persisted until the mid-1990s (Cho, n.d.; Yoon et al., 2020). In the late 1980s, housing prices in Seoul were skyrocketing, causing some serious social concerns. In response, the Korean government embarked on an ambitious initiative to establish new cities in five areas in SMA: Bundang, Ilsan, Pyeongchon, Sanbon, and Jungdong. Phase 1 New Towns focused primarily on housing provision but was not able to meet the desired quality standards for residential areas. This initial effort brought various issues, most notably an imbalanced land use, high property values, and traffic congestion. Acknowledging these issues, Phase 2 New Towns emphasized comprehensive planning and post-development framework that systematically focuses on development equity. The New Towns developed during Phase 2 include ten areas strategically dispersed in SMA (Cho, n.d.; Phase 2 New Town Development Details, 2017; Yoon, 2019). Initiated in 2001, projects for Phase 2 New Towns have made substantial progress toward their objectives.
Even before Phase 2 New Town was completed, New Town projects in Phase 3 began in 2018 and aimed at reinforcing the housing policy stance by providing more housing supplies to solve the ever-increasing real estate prices (Song and Kim, 2019). Therefore, the New Towns in Phase 3 were decided to build near the City of Seoul than those built in Phases 1 and 2 (Gu et al., 2021; Hwang and Kim, 2023). Since the Phase 3 development, New Town projects focused more on the quality of the residential environments and public interests, going beyond the mere objective of increasing housing supply.
Although the planning and design of New Towns require community-level efforts, only a handful studies have tried to document user experiences and their associated outcomes in these areas: Kim (2007) evaluated the planning principles against what’s called “sustainable new town planning criteria” for the cities in the first phase. Yoon (2019) conducted an in-depth expert survey to formulate planning indices and developmental trajectories for Phase 3 New Towns. Additionally, the Land and Housing Corporation (LH) in Korea, the main agency to implement New Town projects, assessed the residents’ quality of life in Phases 1 and 2 New Towns, to understand resident satisfaction of their settlement environment as well as planning components (Yoon et al., 2021). Previous studies have mainly derived results based on indicators such as transportation planning, pedestrian infrastructure, and greenspace ratios from an urban planning perspective. The metrics to evaluate the planning and design outcomes from the development of New Towns are crucial to ensure the success of the development and to inform its future practice. However, the comprehensive evaluation metrics, including user experiences, have not been studied or established, especially for the public space sector including parks and greenspaces as part of new urban fabric. Most research on parks and greenspaces in New Towns focuses on their physical systems. Kim and Choi (2013) found that the consistency between the park greenspace system and the urban street system is closely related to the comfort and sustainability of New Towns. The Phase 1 New Town exhibited less reciprocity between the two, with greenspaces distributed within the city. In contrast, Phase 2 New Town showed relatively high consistency, with qualitative improvements to connect existing greenspace resources. Choi and Jeong (2014) analyzed the types and spatial characteristics of greenspaces as facilities, finding that the type and location of green facilities vary depending on the spatial characteristics of the sites and the land use near the greenspaces, significantly impacting residents’ experiences.
Meanwhile, several studies have explored users’ experiences and perceptions with text data analysis method within Seoul’s prominent neighborhood parks rather than those in New Towns. Woo and Suh (2018) investigated changes in park utilization patterns over time at the Olympic Park. Lee and Son (2021) employed Google Maps reviews to scrutinize user experiences across Seoul Forest, Boramae Park, and Olympic Park. Sa (2024) affirmed public perceptions of urban parks along the Han River, including Seoul Forest, Seonyudo Park, and Yeouido Park. Additionally, Kim (2023) analyzed blog texts focusing on Ttukseom Hangang Park to uncover consumption experience patterns in urban parks.
Given that Phase 1 and 2 New Towns are also developing comprehensive urban environments akin to Seoul, it is imperative to extend such studies to include parks and green spaces within these areas as representative urban parks.
In response to this research gap, this study aimed to analyze park visitors’ experiences in Phase 1 and 2 New Towns with Naver Blog text data quantitatively. The analysis of social media text data is useful for understanding user perceptions and usage patterns through user experience reviews (Kaußen, 2018; Munawir et al., 2019; Song and Zhang, 2020; Huang et al., 2022; Park et al., 2022).
The structural framework of the Phase 3 New Towns district plan, encompassing land use and district unit arrangements, has been officially formulated as of 2023 (Hwang and Kim, 2023). Nonetheless, to create a park that resonates with its potential users, it is important to understand how individuals engage with the park and how they perceive and interact with its attributes. The result of this study could provide a comprehensive landscape planning strategy to improve users’ perceptional engagement with parks and greenspaces during the development of the Phase 3 New Town’s green infrastructure as reflections of previous development.
2. Materials and Methods
The research objects encompass neighborhood parks in two distinct urban locales: Ilsan and Bundang, which are emblematic of Phase 1 New Towns, and Dongtan and Gwanggyo, which are emblematic of Phase 2 New Towns (See Figure 1). Ilsan and Bundang originated in the early 1990s. They conferred approximately three decades of maturation to these localities, rendering consolidated residential domains complemented by an environment characterized by congenial parks and greenspace attributes. Gwanggyo and Dongtan, illustrative of Phase 2 New Towns, were started to build in the 2000s. Phase 2 New Towns formulation systematically infused an expanded network of parks and greenspaces into the urban schema, underpinning a concerted effort to improve the quality of the residential milieu (See Table 1).
Source: Yoon et al., 2020
The selection of parks for the analysis was performed by assessing the number of blog posts ascribed to each neighborhood park within these four cities. The underlying assumption is that highly used places will likely have more posts. Four search terms, including “Ilsan Park”, “Bundang Park”, “Gwanggyo Park”, and “Dongtan Park”, were each queried on Naver, the largest and the most popular search engine and blog service in South Korea (Lee and Son, 2022), to gather blog URLs about the parks from May 20, 2003, marking the initiation of Naver’s service, to May 31, 2022. A series of scraping and filtering methods were applied to acquire complete lists of search results, resulting in 132,328 URLs and metadata, including the post title. Within the lists, the number of blog posts containing the name of each major park was counted, as shown below (See Table 2). The parks with the most posts in each city and their ratios are as follows: Ilsan Lake Park (99.3%), Bundang Yuldong Park (51.8%), Gwanggyo Lake Park (97.8%), and Dongtan Lake Park (52.0%).
Per the site selection process, Ilsan Lake Park, Bundang Yuldong Park, Gwanggyo Lake Park, and Dongtan Lake Park were decided as the research objectives. An overview of these four neighborhood parks is shown below (See Table 3).
| Ilsan Lake park |
Bundang Yuldong Park |
Gwanggyo Lake Park |
Dongtan Lake Park |
|
|---|---|---|---|---|
| Area size | 103ha | 30ha | 17ha | 181ha |
| Completion date | Yr. 1996 | Yr. 1999 | Yr. 2013 | Yr. 2018 |
The retrieval period of text data for the analysis was the same as for the site selection, from May 20, 2003, marking the initiation of Naver’s service, to May 31, 2022. The information extracted included post titles, dates of publication, and body text (docs obtained, ①). Data analysis was performed using the Python programming language.
To enhance the credibility of the collected data, posts associated with real estate and commercial advertising (i.e., “#housing sales”, “#real estate agency”), including duplicates, were eliminated. Moreover, words recurrent in promotional content, such as “#flower shop”, “#hair shop”, and “#auction”, were meticulously identified and omitted if they surfaced even once in the text (docs screened, ②). To isolate posts specifically referencing the park for the analysis from the posts that remained after screening, only those posts containing the specific park names in their titles, namely Ilsan Lake Park, Bundang Yuldong Park, Gwanggyo Lake Park, and Dongtan Lake Park – were selected (docs relevant, ③).
The total number of posts relevant to the parks considered in the analysis was 53,965 (all relevant docs). Subsequently, tokenization was employed for frequency and network analyses. The unique token counts (tokens unique, ④) utilized in the calculation of term frequency–Inverse Document Frequency (TF-IDF) are presented below (See Table 4).
In this study, topics within each park’s corpus were identified and analyzed through keyword extraction using the TF-IDF method. In addition, we employed Word2Vec-based word embeddings to discern the contextual meanings and correlations among keywords within the corpus. TF-IDF contains various formulations tailored for specific analytical purposes and corpus characteristics. In this study, a title-weighted TF-IDF, denoted as twTF-IDF, was utilized. In this approach, text from blog titles is incorporated into the TF-IDF calculation as distinct documents and specific weights are assigned to words emphasized by blog authors as focal points.
For word embeddings, a range of algorithms, from traditional Word2Vec to BERT-based Sentence–BERT, were considered. The resulting vectors in R^d, which represent a matrix of several words derived using Word2Vec, encapsulate the compressed semantic space within the d dimensions of the corpus. A dimension-reduction algorithm is applied to visualize the space. Notably, after experimenting with well-known techniques such as Principal Component Analysis (PCA), t-distributed Stochastic Neighbor Embedding (t-SNE), and Multidimensional Scaling (MDS) for 2D compression, MDS demonstrated the best fit for the objectives of this study.
3. Results
To investigate people’s utilization and perception of parks within Phase 1 and 2 New Towns, a keyword frequency analysis (N=50) was conducted for each park, yielding valuable insights. Table 5 lists the 20 keywords with the highest validity, and Figure 2 shows the percentage of the text group with the same characteristics.
The findings revealed that all four parks were used for relaxation and recreation. Notably, “strolling” and “outing” emerged as the most frequently used words in the twTD-IDF frequency analysis, ranking in the top three in all four parks. In Ilsan Lake Park and Bundang Yuldong Park, “strolling” ranked 1st and 2nd, respectively, while “outing” ranked in the opposite direction. In Gwanggyo Lake Park and Dongtan Lake Park, “strolling” consistently held first, while “outing” ranked 3rd in both parks.
Situated within New Town areas, these parks were categorized as “neighborhood parks” under the “Act on Urban Parks and Green Areas (2016)” in Korea. While the four New Town parks are relatively substantial at over 50 ha (except Bundang Yuldong Park), they operate as daily greenspaces within residential locales, diverging from the concept of large theme parks. Apart from “strolling” and “outing,” common park activities encompassed “bicycle”, “exercise”, “walking”, and “picnic”. Importantly, posts related to the park experience during “spring” and “autumn” were prevalent and landscape elements such as “flower” and “cherry blossom” came up in all four parks, indicating the significance of the “Landscape” factor in shaping the park usage experience.
Second, the analysis underscores park’s’ diverse user groups. All four parks were conveniently situated near residential areas, ensuring accessibility for the general public. Notably, “kids” and/or “family” consistently rank within the top 20 keywords across all four parks in frequency analysis except Ilsan Lake Park. This prevalence indicates park’s appeal to family-oriented user groups, aligning with the overarching demographic focus of Phase 1 and 2 New Towns, which primarily cater to middle-class and young families.
Interestingly, in Bundang Yuldong Park, the phrase “hanging out with lovers” holds similar rankings (7th) compared to “kids” (6th) and much higher than “family” (16th). This observation suggests a notable emphasis on recreational activities among couples visiting parks. Moreover, neighborhood parks in Phase 1 and 2 New Towns serve as versatile spaces catering to various user groups. In addition to their family-centric leisure and relaxation functions, these parks operate as multipurpose hubs that foster interactions with surrounding commercial districts.
Third, the programs for parks nearby were also related to park usage. In Bundang Yuldong Park and Dongtan Lake Park, a significant portion of the high-frequency words pertained to park external elements (abbreviated as [C]), often accompanied by the term ‘famous restaurant’ and/or ‘shopping mall,’ representing external elements within the park’s vicinity. Specifically, Bundang Yuldong Park accounted for 12.2% and Dongtan Lake Park accounted for 12.4%.
Fourth, compared to the frequently occurring terms “strolling” and “outing”, the words within the top 20 rankings represented distinctive park elements or content/programs specific to each park. For instance, at Ilsan Lake Park and Bundang Yuldong Park, the terms “International Flower Festival”, “Bungee Jumping”, “and” “Book Theme Park” ranked high. In Gwanggyo Lake Park, “Shindae Lake” and “Woncheon Lake”, “the original lakes in the park, were ranked within the top 10, along with the park’s unique programs such as” “campsite”. “In Dongtan Lake Park, “Luna Show” and “Lake Como”, the shopping mall, appear in the second and fifth positions, respectively.
These words, pertaining to each park’s exclusive elements and content/programs, establish a robust network that constitutes the park’s unique characteristics that are discernible to visitors. The recognition of these characteristics was further substantiated through network analysis, as discussed in the subsequent section.
Through the network, distinct user scenarios within the park were identified based on representative keywords that illuminated the park’s essence. This network was visualized as a 50-word (token) network derived from twTF-IDF, with the font size of each token proportional to its twTF-IDF value. In this network, an intertoken edge was drawn when the cosine distance of the word embeddings fell within the top 200 (approximately 0.1% of the total), with the width of the edge representing the relative similarity.
The results of the network analysis delineated four overarching types of park users. The first group (G1) comprised individuals engaging in everyday leisure activities within the park, including strolling and outings, which describes an experience centered on user behavior. The second group (G2) encompassed users attracted to park events and content, such as fairs and festivals, located around the purpose of the user’s visit. The third group (G3) comprised individuals who appreciated and enjoyed the natural elements, plants and scenic views of the park, an essential aspect from a research point of view as it relates to the natural things of the park. The last group (G4) primarily utilized nearby commercial facilities, including restaurants, which focused on the use of external facilities.
Specifically, Ilsan Lake Park exhibited characteristics primarily aligned with the G2 type, emphasizing the “International Flower Festival (IFF)”. Related terms like “event”, “exhibition”, and “festival” formed a cohesive network within this group. The park’s thematic focus on flowers and plants was evident through the inclusion of various plant species such as “rose”, “cherry” and “tulip”. The G3 group, appreciative of nature and landscapes (such as “flower viewing”, “scenery”, “sunset”,” and “sundown”), formed distinct clusters in association with the G2 type (See Figure 3). The G1 group, centered around daily leisure activities (such as “bicycle”,” “exercise”,” and “strolling”), was also shown clearly.
While Ilsan Lake Park was characterized by a prominent focus on content and programs, indicating a strong presence of G2 type users, Gwanggyo Lake Park mainly featured a diversity of leisure activities within the park, aligning with G1 type. Gwanggyo Lake Park was characterized by its comparatively low proportion of content and programs. Instead, the park’s main perceived features among park-goers were its park elements (“camp site”, “view tower”) and specific locations (“Woncheon Lake” and “Shindae Lake”, existing lakes within the park boundary). The G1 type words in Gwanggyo Lake Park spanned a wide range, including “bicycle”, “photo trip”, “excursion”, “camping”, and “walking”. However, these terms exhibited low interrelations and did not form discernible cohesive groups. In Gwanggyo Lake Park, networks by the location – “night view” at “Shindae Lake”, “view tower” at “Woncheon Lake” and “outing” at “campsite” – were observed (See Figure 4).
Compared with the other two parks, Bundang Yuldong Park and Dongtan Lake Park exhibited weak and fragmented networks. Both parks display a scattered arrangement of terms, with isolated instances like “bungee jumping” in Bundang and “Luna (water show)” in Dongtan standing out. Moreover, the prevalence of keyword [C], which indicates external commercial facilities, signified a dominant presence in these areas. However, a notable disparity emerged between Bundang and Dongtan concerning the relationship between the utilization of nearby commercial facilities (G4) and park user experience.
Despite their weak connections and distinct characteristics, Dongtan Lake Park operates as an integrated space where the park and surrounding commercial facility areas, as perceived by park users, function as interconnected entities (See Figure 5). By contrast, Bundang Yuldong Park portrays a disjointed scenario. Despite the exclusion of numerous blog advertisements, content related to commercial facilities overshadowed the park’s experience. In the case of Bundang Yuldong Park, user experience was fragmented, presenting the park as a passive peripheral space primarily associated with commercial activities (See Figure 6).
Bundang Yuldong Park and Dongtan Lake Park show a clearer network between G1, G2, G3 and G4 than the previous two parks. First, restaurants around the park (G4), that is, programs outside the park were connected to “hanging out with lovers”, which in turn are linked to activities such as “outing”, “strolling”, ”“walking around”,“ and “exercise” (G1). Another thing is that enjoying the park’s natural elements in all four seasons was strongly expressed, with “pink muhly”, “reed” and “cherry blossom” centered around the park’s autumn scenery (G3). Lastly, the “bungee jump tower” and “book-themed park”, facilities and content/programs unique to Bundang Yuldong park, appeared and were interconnected (G2) (See Figure 5).
Dongtan Lake Park displayed daily activities, such as “strolling”, “outing”, “exercise” and “picnic.” (G1). Also, Dongtan Lake Park’s unique content/programs centered around the Luna Show - “music fountain”, “media show” and “night view“” appeared (G2). The scenery that allows people to feel the park’s seasonal landscape, such as “reed” and “cherry blossom”, was shown as a strong network (G3). “Lake Como”, the shopping mall near the park (G4), was prominent but had weak connections to other aspects of the park. However, it is understood that it plays a significant role in park users recognizing the park.
4. Discussion
The user experiences of the four New Town parks revealed the originality of the city park’s attributes and functions. Since the inception of modern citizen parks, urban parks have been envisioned as public nature gardens offering spaces where individuals can connect with nature, indulge in leisure activities and foster familial bonds (Hwang, 2014; Sa, 2024). Even in contemporary society, parks persist as venues for “strolling” and “outing”, and serve as everyday spaces for family engagement (Huang et al., 2022; Sreetheran, 2017). Terms such as “picnic”, “bicycle”, “scenery”, and “tree” within park contexts serve to validate the enduring significance of rest, leisure and natural elements in these spaces, regardless of changing times, thereby enriching the quality of urban life (Lee and Son, 2021; Kim, 2023).
Beyond these park characteristics, the unique park elements and distinctive content/programs specific to each park enhance public awareness and diversify individual park experiences. For instance, Ilsan Lake Park focuses on the International Flower Fair and shapes park elements and content/programs accordingly. Bundang Yuldong Park has bungee jumping and library (book-themed park) facilities. Dongtan Lake Park has a clear group of park elements – a music fountain for the Luna Show. In contrast, Gwanggyo Lake Park, featuring fewer content/programs, is strategically positioned to foster daily leisure as a cultural facet of life in its proximity to residential areas. This strategic approach integrates various park elements with inherent natural elements, such as Woncheon and Shindae lakes, into the park’s development plan (Kim et al., 2009).
Thus, a park’s identity evolves through public perceptions and usage patterns, forming multiple layers influenced by its inherent nature, facilities, content and programs. A park’s identity becomes profoundly ingrained in the public consciousness when it hosts distinctive programs while preserving its fundamental naturalistic characteristics and ensuring that it remains unique and recognizable in the future.
With the global proliferation of social media usage, extensive research on text-mining analysis is actively underway to comprehend user experiences and perceptions within this domain. This trend is primarily driven by enhanced information accessibility due to technological advancements, enabling a comprehensive exploration of public interest and perceptions (Park et al., 2022; Yang and Liu, 2022). The use of social media big data for text mining research in public spaces, including urban parks, enables the detection of user perceptions and usage patterns through user experience reviews (Huang et al., 2022; Kaußen, 2018; Munawir et al., 2019; Park et al., 2022; Song and Zhang, 2020). These user-generated narratives enhance the objectivity of the research regarding the compilation of user opinions when juxtaposed with traditional evaluation results obtained through conventional surveys (Bernabeu-Bautista et al., 2023; Choi et al., 2022; Koblet and Purves, 2020; Sim and Miller, 2019).
While social media text analysis is valuable for discerning trends, perceptions and behavioral shifts within extensive datasets, many studies simultaneously consider the geospatial context or land use of specific public spaces (Bernabeu-Bautista et al., 2023; García-Palomares et al., 2018). Understanding perceptions and user activities within a specific park context necessitates consideration of the park’s physical environment. In particular, examining the characteristics of areas adjacent to parks is crucial for comprehending and interpreting people’s perceptions and behavioral patterns evident in social media texts.
The text analysis result in this study, for example, shows that adjacent commercial facilities such as restaurants and shopping malls are privileged texts for Bundang Yuldong park and Dongtan lake park. This indicates that nearby commercial facilities would be one of the drivers of people using the park.
Therefore, although text mining offers the advantage of efficiently analyzing vast amounts of data using computer algorithms, understanding the geographical features and contextual elements surrounding a park’s physical environment is essential. This contextual understanding enhances the accuracy of interpreting the revealed phenomena. A precise interpretation of these results can inform park planning and design and guide the creation of urban public spaces that support New Town park users’ daily lives and leisure cultures.
5. Conclusions
This study focused on four representative parks in Phase 1 and 2 New Towns and encompassed deep learning-based Naver blog text analysis, finding visitors’ activities and perceptions to derive insights into park and greenspace planning and to establish strategies for Phase 3 New Towns. This study aims to understand how people perceive and utilize these parks. The findings revealed that all four new town parks exhibited usage patterns oriented toward daily relaxation and leisure activities. These patterns attract diverse users, including residents and other groups. Furthermore, the uniqueness of each park was shaped not only by its facilities but also by its distinctive content and programs. The findings of this study provide valuable insights for planners, landscape architects and policymakers involved in the planning and development of parks and greenspaces in Phase 3 New Towns. Rather than designing parks as an afterthought once urban planning has been finalized, this study underscores the importance of integrating park development considerations into the initial stages of urban planning. Parks should be designed to offer unique content while preserving their natural characteristics.
The results of the present study have several important implications. First, it underscores the fact that the representatives of a park evolve through multi-layered processes involving people’s perceptions, park usage patterns and the interplay between the park’s inherent attributes and its elements, content and programs. Second, it emphasizes that a comprehensive approach to urban planning encompassing the park and its surrounding areas can extend the scope of how it is perceived and utilized. Such integrated planning can enhance awareness and usage of parks. Finally, besides text data analysis, this study discusses and underscores the potential value of incorporating analyses for physical spaces and broader social environments as complementary methods to aid more accurate interpretations of text data.
There are some limitations to this study. First, while it aimed to understand how people perceive and use these four parks as neighborhood parks, they attract not only local residents but also a significant number of transient visitors. This makes it challenging to consider the opinions exclusively representative of the residents. The study also suggests that park usage patterns vary based on the physical characteristics of the parks and the surrounding land use. However, it did not statistically examine the correlation between geospatial and text analysis results. Conducting these two analyses together would provide a more in-depth understanding of the usage patterns and user experiences in new town parks, leading to more accurate planning and design implications for Phase 3 New Towns.
