Application of a DPSIR-Based Causal Framework for Sustainable Urban Riparian Forests: Insights from Text Mining and a Case Study in Seoul

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The study constructs a causal model based on the DPSIR framework, systematically applying text mining and network analysis techniques to the field of urban riparian forest management. It breaks through the limitations of traditional single-ecological perspectives and proposes a multi-dimensional framework integrating environmental, social, and policy factors for urban riparian forest management. The combination of the DPSIR framework with text mining (SciBERT clustering) and spatial analysis reveals the complex causal chains in urban riparian forest management. Notably, the validation of the correlation among water quality, land use, and biodiversity demonstrates academic value. This research is highly interesting, methodologically innovative, and practically applicable.

The manuscript requires further improvement in the following aspects:

1.The text mining section mentions "100 nodes and 322 edges" but does not specify the screening criteria; the case study abruptly introduces the "30 Million Tree Planting Program" without explaining its relevance to riparian forests. Such leaps affect reproducibility. It is recommended to supplement flowcharts and data screening criteria.

2.There are some data presentation and visualization issues, for example, Figure 3 (Visualizing the Relationships between Key Concepts) has excessive nodes, reducing readability, and fails to highlight core pathways; the "Other" category in Table 2 (DPSIR classification) occupies a large proportion, weakening the framework's explanatory power.

3.In subsection 4.3, the authors found a positive correlation between population density and bird abundance, attributing it to "green gentrification," but did not rule out confounding factors, such as observational bias caused by more monitoring points in high-density areas. Controlled variable analysis or spatial regression validation should be added.

Author Response

Comment 1:
The text mining section mentions "100 nodes and 322 edges" but does not specify the screening criteria; the case study abruptly introduces the "30 Million Tree Planting Program" without explaining its relevance to riparian forests. Such leaps affect reproducibility. It is recommended to supplement flowcharts and data screening criteria.

Response:
We greatly appreciate the reviewer’s valuable comment highlighting the need for clearer explanation of our screening criteria and case study rationale. In response, we have revised the manuscript in two key aspects:

1. Clarification of keyword filtering criteria in the text mining process (Section 2.2.1):
   We have now explicitly described the threshold used for keyword inclusion in the co-occurrence network. Specifically, only keywords that appeared at least 15 times across the corpus and co-occurred with at least one other keyword that also appeared at least 15 times were retained. This filtering was applied to improve network clarity by focusing on meaningful, frequently discussed concepts and reducing noise from rare or isolated terms. Based on this criterion, we selected the top 100 keywords to construct the network. The revised sentence is shown below.
   
   ⧠ Revised Sentence in Section 2.2.1: The selected documents underwent keyword co-occurrence network analysis and SciBERT-based sentence clustering. To improve network clarity, we applied a filtering threshold in which only keywords that appeared at least 15 times across the corpus and co-occurred with at least one other keyword that appeared at least 15 times were retained. Based on this criterion, the top 100 keywords were selected to construct the co-occurrence network, which consisted of 100 nodes and 322 edges, with a network density of 0.065. Modularity analysis using Gephi (v0.10.1) and the ForceAtlas2 layout algorithm revealed six modular clusters, yielding a modularity score of 0.349.
   
   
2. Justification of the “30 Million Tree Planting” initiative in relation to riparian forest management (Section 2.1):
   We have revised the text to emphasize that this large-scale urban afforestation program directly targets waterfront and riparian areas, which are core zones for buffer enhancement, ecological restoration, and carbon sequestration. This strengthens its relevance to the DPSIR framework’s Response The revised paragraph is shown below.
   ⧠ Revised Paragraph in Section 2.1: The selection of these streams aligns with Seoul’s “30 Million Tree Planting” initiative, a large-scale afforestation policy designed to expand urban green spaces, enhance waterfront and riparian buffer zones, and support the city’s carbon neutrality goals [8]. This initiative specifically targets waterfront and riparian areas as strategic sites for increasing ecological connectivity, improving air and water quality, and mitigating the urban heat island effect. To ensure accurate ecological observations, this study concentrated on riparian zones located beyond the levees—areas that are less affected by artificial infrastructure and more representative of natural ecological processes. These zones, characterized by native vegetation and limited human disturbance, play a central role in Seoul’s climate change mitigation efforts. They also serve as critical sites for evaluating the ecological functions of urban riparian forests and provide essential ecosystem services, including water purification, flood regulation, carbon sequestration, and wildlife habitat provision.

We believe these modifications substantially improve the reproducibility and transparency of our methodology and appreciate your thoughtful suggestions.

 

Comment 2:
There are some data presentation and visualization issues, for example, Figure 3 (Visualizing the Relationships between Key Concepts) has excessive nodes, reducing readability, and fails to highlight core pathways; the "Other" category in Table 2 (DPSIR classification) occupies a large proportion, weakening the framework's explanatory power.

Response:
We appreciate the reviewer’s thoughtful feedback regarding the clarity of Figure 3 and the interpretability of Table 2. We fully agree that both visual and tabular elements are critical to effectively communicating the framework. Accordingly, we would like to provide clarification on the intent and rationale behind our current presentation:

1. Regarding Figure 3 (Visualizing the Relationships between Key Concepts):
   We acknowledge the reviewer’s concern about the high number of nodes in Figure 3. However, we would like to clarify that the current version reflects the result of an extensive optimization process. We attempted multiple configurations and filtering levels during the network construction phase and concluded that this version offered the best balance between visual readability and thematic interpretability. While reducing the number of displayed nodes might improve clarity, it would simultaneously compromise the modular structure and weaken the analytical value of the cluster analysis, which relies on sufficient density to detect thematic groupings. For this reason, we maintained the current structure.
   
   
2. Regarding the “Other” category in Table 2
   Thank you for pointing out the ambiguity of the “Other” category in our DPSIR classification table. In response, we revised the label from “Other” to “Excluded” to clarify that the terms in this category were not ignored but rather intentionally omitted from DPSIR mapping due to lack of conceptual fit.
   We would also like to clarify that this classification step was part of a larger iterative process inherent to text mining and conceptual framework development, where many terms are repeatedly filtered and categorized to enhance thematic focus. To reflect this intent more clearly, we added an explanatory note directly beneath Table 2.
   
   ⧠ Revised Note under Table 2:
   Note: The “Excluded” category refers to high-frequency keywords that were intentionally omitted from the DPSIR classification due to a lack of direct conceptual alignment with the framework’s components.

 

Comment 3:
In subsection 4.3, the authors found a positive correlation between population density and bird abundance, attributing it to "green gentrification," but did not rule out confounding factors, such as observational bias caused by more monitoring points in high-density areas. Controlled variable analysis or spatial regression validation should be added.

Response:
We thank the reviewer for this important observation regarding the potential influence of confounding variables in interpreting the correlation between population density and bird species richness. We acknowledge that observational bias—such as increased monitoring activity or data availability in densely populated urban areas—may partly account for the observed relationship.

In response, we have revised Section 4.3 to explicitly state this analytical limitation and to present the interpretation of green gentrification as a plausible, yet tentative, hypothesis. At the same time, we would like to emphasize that recent demographic trends in Seoul support the relevance of this concept. Specifically, some riverside districts in Seoul are experiencing population decline despite substantial ecological and infrastructural improvements, a pattern that aligns with the defining characteristics of green gentrification as discussed in the literature.

To reflect this balanced perspective, we have also revised Section 4.4 to recommend that future research apply spatial regression or multilevel modeling techniques that account for observation effort, land-use heterogeneity, and socio-economic variables. These adjustments strengthen the validity of our conclusions and outline a path for more robust causal analysis in subsequent studies.

 
⧠ Revised Sentence in Section 4.3:

Unlike prior studies that primarily focused on ecological and environmental variables, this study emphasizes the importance of incorporating social and economic dimensions into urban riparian forest management. One notable finding is the positive correlation observed between population density and bird species richness. While this may suggest that denser areas with more green infrastructure support higher biodiversity, we acknowledge that this pattern could be partially influenced by observation bias, such as increased monitoring activities or citizen science participation in more urbanized regions.

However, beyond statistical interpretation, recent empirical trends in Seoul support the relevance of the green gentrification hypothesis. Several riverside districts—despite benefiting from ecological and infrastructural improvements—are experiencing population decline, a trend often associated with green gentrification, where environmental enhancement contributes to changes in the social or demographic composition of neighborhoods. In this light, our interpretation of green gentrification is presented as a contextually grounded and plausible mechanism, though further research with spatially controlled models is necessary to validate this hypothesis.

⧠ Revised Sentence in Section 4.4: While the DPSIR framework applied in this study provides a useful structure for identifying environmental drivers and pressures, it has limited capacity to capture the complexity of interactions between urbanization, climate change, and socio-economic processes. In particular, factors such as green gentrification, engineered water infrastructure, and urban governance dynamics remain underrepresented in the conventional DPSIR approach. Moreover, the study relied primarily on publicly available datasets collected by administrative and research institutions. Although these datasets offer consistent and city-wide coverage, they are often constrained by fixed monitoring locations, limited temporal resolution, and uneven observation intensity across spatial units. Such characteristics may influence correlation-based findings, particularly in urban areas with varying levels of data density and reporting activity.

However, the case of Seoul provides an important empirical context where these limitations become analytically valuable. For instance, despite improvements in ecological conditions along urban waterways, some riverside districts have experienced population decline—demonstrating a real-world pattern consistent with green gentrification dynamics. This suggests that integrating socio-economic dimensions is not merely theoretical but necessary for understanding contemporary urban ecological transitions.

Future research should therefore aim to develop a multidimensional modeling framework that incorporates social dynamics, economic behavior, and feedback-based mechanisms. AI-driven tools such as system dynamics modeling (SDM), agent-based modeling (ABM), and spatial causal inference approaches may provide more accurate simulations and stronger policy guidance. Furthermore, long-term data on biodiversity, land-use, and civic engagement will be essential for monitoring the effectiveness of riparian forest interventions. Ultimately, moving toward an integrated framework that unites environmental, social, and economic perspectives will better support evidence-based policymaking for resilient, equitable, and ecologically functional urban landscapes [22, 24, 39].

Reviewer 2 Report

Comments and Suggestions for Authors

“A DPSIR-Based Causal Framework for Sustainable Management of Urban Riparian Forests: Insights from Text Mining and a Case Study in Seoul” is quite an interesting manuscript. Unfortunately, I do not understand how the authors arrive at the correlations shown in Figure 4. This needs to be (better) explained. More things in the following notes.

 

TITLE

In my opinion, the term “management” is not absolutely necessary in the title. It may be omitted.

 

 

INTRODUCTION

Page 1, first sentence: For the global statement (“human societies and urban settlements have emerged along riverbanks”), a less regional but more general reference should also be possible.

 

Page 2, fifth line: Better “Such changes have reduced habitat quality...”.

 

Page 2, line 14 (second paragraph): Perhaps “habitat provision” instead of “creation”?

 

Page 2, third paragraph: Better 'invasive pecies spread'

 

 

METHODOLOGY

Page 4, line 6 and following: What are the PRISMA protocol and SciBERT-based sentence classification? The former is only written out in its long form on page 5.

 

Page 5, Figure 2: The size of the input data set (1,001 research abstracts) is repeatedly mentioned; can Step 5 also be summarised numerically?

 

Page 5, line 6: In the manuscript it is not consistent until "June 2023" - In the abstract and in 2.2. Methods (page 4) it says ‘academic journal articles and reports published between January 2010 and June 2024’!!

 

Page 5, second paragraph: How could the reviewers be ‘independent’?

 

Page 6, sixth paragraph: The ‘water quality parameters’ should also be written out first (BOD, T-P).

 

 

RESULTS

Page 7, subheading 3.1.: Either ‘Development of a causal framework based on the DP-SIR framework through text mining’ or ‘Developing a causal framework based on the DP SIR framework through text mining’.

 

Page 8, Figure 3: Some of the cluster terms already appear in the figure (such as water quality and land use), others do not. The traceability of the assignment via the colours is helpful here.

Page 9, line 2: The spelling is ‘season’.

 

Page 9, third paragraph: What about increasing stochasticity, for example in connection with climate change? I am surprised that no corresponding term appears in the context of the term ‘climate change’. For example, low water (levels) in rivers that cause ‘new’ major problems, such as an increase in salt levels on the River Oder in connection with waste water from paper mills in Poland. This has led to enormous fish mortality (https://www.nature.com/articles/s41598-024-66943-9). Temporary or episodic rivers already exist in Central Europe, and especially in regions of the world where less is publicised. You should at least roughly outline this in your spectrum framework or consider it in the outlook (the problems of increasing low water already exist in significant regions of the world, but there seems to be hardly any scientific literature on it yet). To get some ideas and background here, I recommend that you use the following publication (https://doi.org/10.1111/aje.12820).

 

Page 9 and following, Table 1: To save space, landscape format should be selected.

 

Page 10, Table 1, Cluster ID 1, second bullet point: 0.01 to 0.06 km²

 

Page 10, Table 1, Cluster ID 1, third bullet point: ecosystem classification + principles

 

Page 10, Table 1, Cluster ID 1, fourth bullet point: regional

 

Page 10, Table 1, Cluster ID 1, fifth bullet point: designation

 

Page 10, Table 1, Cluster ID 2, column "Top 10 Words": CO₂

 

Page 14, line 1: Should it really start with ‘|’? Why?

 

Page 15, line 3: ecological

 

Page 16, line 3: BOD (Biochemical Oxygen Demand). This should be written out in full the first time BOD is mentioned in the manuscript text!

 

Page 18, Figure 5: It seems that with ‘Population Density’ only a single ‘Driving Force’ has been recognised, so I recommend changing it to the singular in the legend.

 

Page 19, second paragraph: Additionally

 

 

DISCUSSION

Page 20, subheading 4.1: Perhaps ‘Identification of key themes in urban riparian forest management through text mining and network analysis’?

 

Page 20, penultimate line: acknowledge

 

Page 21, subheading 4.2.: Urban

 

Page 21, second paragraph: health [33], the analy-

 

Page 21, second paragraph: highlights

 

Page 22, line 3: allow

 

Page 22, second paragraph: surface

 

Page 22, fifth paragraph: addressing + address

Author Response

Author's Response to Reviewer 2

We thank Reviewer 2 for the thorough and insightful feedback. Below, we address each comment using the standard format.

 

Comment 1: In my opinion, the term “management” is not absolutely necessary in the title. It may be omitted.

Response 1: Thank you for pointing this out. We agree with this comment. Therefore, we have removed the word “management” from the title to enhance conciseness without altering the conceptual focus of the study.

[Revised title: “Application of A DPSIR-Based Causal Framework for Sustainable Urban Riparian Forests: Insights from Text Mining and a Case Study in Seoul” – updated in manuscript title and header.]

 

Comment 2: Page 1, first sentence: For the global statement (“human societies and urban settlements have emerged along riverbanks”), a less regional but more general reference should also be possible.

Response 2: Thank you for pointing this out. We agree with this comment in principle. However, we did not revise the sentence because we were unable to identify a more appropriate and comprehensive reference that would better reflect global settlement patterns along riverbanks than the current citation. We acknowledge the importance of this suggestion and will consider including a broader reference in future work.

 

Comment 3: Page 2, fifth line: Better “Such changes have reduced habitat quality...”

Response 3: Thank you for your suggestion. We revised the sentence accordingly.
[Page 2, Introduction, paragraph 1, line 5: "Such changes have reduced habitat quality..." now used.]

 

Comment 4: Page 2, line 14: Perhaps “habitat provision” instead of “creation”?

Response 4: Thank you. We have revised the wording to "habitat provision" as suggested.
[Page 2, Introduction, paragraph 2.]

 

Comment 5: Page 2, third paragraph: Better “invasive species spread”

Response 5: Revised. We replaced “introduction of invasive species” with “spread of invasive species”.
[Page 2, paragraph 3.]

 

Comment 6: Page 4, line 6 and following: What are the PRISMA protocol and SciBERT-based sentence classification?

Response 6: Thank you for this suggestion. We have added brief definitions for the PRISMA protocol and SciBERT-based classification early in Section 2.2.
[Page 4, Section 2.2, paragraph 1.]

 

Comment 7: Page 5, line 6: Inconsistency in date – “June 2023” vs. “June 2024”

Response 7: Thank you. We corrected all occurrences to consistently state “June 2024.”
[Page 4–5, multiple locations.]

 

Comment 8: Page 5, second paragraph: How could the reviewers be “independent”?

Response 8: We clarified this by stating that “three members of the research team reviewed abstracts to minimize bias.”
[Page 5, Section 2.2.1.]

 

Comment 9: Page 6, sixth paragraph: Write out “BOD, T-P” in full

Response 9: We revised the manuscript to provide full terms at first mention: “Biochemical Oxygen Demand (BOD)” and “Total Phosphorus (T-P)”.
[Page 6, Section 2.2.2.]

 

Comment 10: Page 7, subheading 3.1: Better consistency in subheading wording.

Response 10: Revised to: “Development of a causal framework based on the DPSIR framework through text mining.”
[Page 7, Subheading 3.1.]

 

Comment 11: Page 8, Figure 3: Improve traceability of cluster assignments.

Response 11: Thank you for the suggestion. We explored several options to enhance the traceability of cluster assignments in Figure 3. Despite multiple attempts, the current version represents the best possible visualization given the structure of the data and layout constraints. While the color-coding has been improved and key terms made more explicit, we acknowledge that some limitations remain in depicting all cluster terms simultaneously.

[Page 8, Figure 3.]

 

Comment 13: Page 9, line 2: Typo – “season”

Response 13: Corrected the spelling.
[Page 9, Section 3.1.2.]

 

Comment 14: Page 9, third paragraph: Add discussion on stochasticity (climate-related low water).

Response 14: Thank you. We added a new paragraph discussing climate-induced stochasticity and low-flow events, including reference to the Oder River incident and related literature.
[Page 22, Section 4.3, paragraph 4.]
“In addition to these socio-ecological considerations, our study acknowledges an emerging yet underexplored dimension in urban riparian research: the increasing hydrological stochasticity associated with climate change. While our text mining results identified frequent mentions of “climate change,” they did not adequately capture its connection to extreme or episodic hydrological events, such as prolonged low-flow conditions, flash droughts, and temporary rivers. Recent environmental incidents—such as the mass fish mortality in the Oder River caused by elevated salinity levels during a period of exceptionally low discharge—highlight the urgency of addressing these dynamics in both research and practice [39]. Such events are no longer rare anomalies but are becoming recurring stressors in riverine systems across Central Europe and beyond [39, 40]. Given the current lack of scientific literature on this subject, incorporating the concept of hydrological variability and stochastic disturbance into future iterations of the DPSIR framework could improve its sensitivity to climate-induced risks.”

 

Comment 15–19: Various wording corrections in Table 1 (e.g., “0.01–0.06 km²”, “ecosystem classification”, “designation”, “CO2”)

Response 15-19: All wording corrections have been made as suggested.
[Page 10, Table 1.]

 

Comment 20: Page 14, line 1: Remove stray “|” symbol

Response 20: Corrected. The symbol has been removed.
[Page 14.]

 

Comment 21: Page 15, line 3: Typo – “ecological”

Response 21: Corrected.
[Page 15.]

 

Comment 22: Page 16, line 3: First mention of BOD should be written in full.

Response 22: Already addressed in Section 2.2.2.

 

Comment 23: Page 18, Figure 5: “Population Density” – use singular.

Response 23: Revised to “Driving Force”
[Page 18-20, Figure 5-6 legend.]

 

Comment 24: Page 19, second paragraph: Add missing word “Additionally”

Response 24: Corrected.
[Page 19, paragraph 2.]

 

Comment 25: Page 20, subheading 4.1: Improve clarity of heading.

Response 25: Revised to: “Identification of key themes in urban riparian forest management through text mining and network analysis.”

 

Comment 26–31: Various minor word corrections (e.g., “acknowledge,” “highlights,” “health,” “analy-,” “allow,” “surface,” “addressing,” “address”)

Response 26-31: All typographical and grammatical corrections have been made.
[Pages 20–22.]

 

We hope these revisions meet your expectations and thank you again for your valuable and constructive comments.

Reviewer 3 Report

Comments and Suggestions for Authors

Work well done,  and paper well-written, so much that I almost didn't know how to review it. However, I recommend some adjustments to be made before consideration of the paper for publication (see attached pdf file with annotations). An example of a major concern is the results being presented in a longish manner - see comments.

Comments for author File: Comments.pdf

Author Response

Author's Response to Reviewer 3

Manuscript Title: Application of a DPSIR-Based Causal Framework for Sustainable Urban Riparian Forests: Insights from Text Mining and a Case Study in Seoul
Manuscript ID: forests-3731352
Journal: Forests (MDPI)

 

Dear Reviewer,

We would like to sincerely thank you for your thorough and constructive feedback on our manuscript. We appreciate your positive evaluation as well as the insightful comments that helped us improve the clarity, focus, and academic rigor of our work. Below, we provide a detailed point-by-point response to each of your 14 comments. All corresponding changes have been incorporated into the revised manuscript and highlighted accordingly.

We hope that the revisions meet your expectations and that the manuscript is now suitable for publication. Thank you again for your valuable contribution to enhancing the quality of our research.

 

Comment 1: Title should include 'application' of DPSIR

Response: Changed to highlight the applied nature of the study and remove redundancy in the phrase 'sustainable management'.

• Before: "A DPSIR-Based Causal Framework for Sustainable Management of Urban Riparian Forests: Insights from Text Mining and a Case Study in Seoul"
• After: "Application of a DPSIR-Based Causal Framework for Sustainable Urban Riparian Forests: Insights from Text Mining and a Case Study in Seoul"

 

Comment 2: Rewrite abstract based on 6 comments

⧠ Comment 2-1. “It would be more stronger finding if there was ground-truthing comment, did you find the same on the ground of the study site?”

Response:

We appreciate this comment. While the framework was primarily developed through text mining of secondary literature, we emphasized its empirical grounding by applying it to four actual riparian sites in Seoul. Within the revised abstract, we now highlight that the framework was validated through indicator-based analysis and correlation mapping across these real-world sites, thereby addressing the importance of on-the-ground application and relevance.

 

⧠ Comment 2-2. “Good comment, but it must address six themes identified – see my previous comment.”

Response:

We have revised the abstract to explicitly list the six major thematic clusters identified through text mining: water quality, ecosystem services, basin and land use management, climate-related stressors, anthropogenic impacts, and greenhouse gas emissions. This correction improves clarity and ensures that the abstract better reflects the central findings of the thematic analysis.

 

⧠ Comment 2-3. “What does this expert validation entail?”

Response:

While the abstract space is limited, we clarified the empirical validation by noting the application to Seoul and the use of measurable indicators. In the main text, we detail that five external experts in urban ecology and riparian forestry were involved in reviewing the DPSIR classification and relationships. Their input helped ensure the thematic structure's relevance and coherence.

 

⧠ Comment 2-4. “Only forests or overall environmental management then examples being forest?? You can rephrase.”

Response:

We revised the abstract to indicate that the model, though constructed in the context of urban riparian forests, is transferable to other urban ecological contexts. This change reflects the broader applicability of the methodological approach beyond riparian forests alone.

 

⧠ Comment 2-5. “Add the relevance of effective management under climate change uncertainties – I recommend to increase applicability of the results or model.”

Response:

Thank you for this suggestion. We revised the final sentence of the abstract to directly emphasize that the proposed model supports adaptive planning and evidence-based decision-making under the uncertainties posed by climate change, thereby addressing both relevance and transferability.

 

⧠ Comment 2-6. “Questions well framed – pls reflect this in the abstract also.”

Response:

We appreciate the recognition of the structured research questions. The revised abstract now incorporates this framing by clearly stating the problem, outlining the method (text mining + DPSIR), presenting the analytical themes, and describing both the theoretical structure and its real-world validation.

⧠ A final revision of abstract
As urbanization accelerates and climate change intensifies, the ecological integrity of urban riparian forests faces growing threats, underscoring the need for a systematic framework to guide their sustainable management. To address this gap, we developed a causal framework by applying text mining and sentence classification to 1,001 abstracts from previous studies, structured within the DPSIR (Driver–Pressure–State–Impact–Response) model. The analysis identified six dominant thematic clusters—water quality, ecosystem services, basin and land use management, climate-related stressors, anthropogenic impacts, and greenhouse gas emissions—which reflect the multifaceted concerns surrounding urban riparian forest research. These themes were synthesized into a structured causal model that illustrates how urbanization, land use, and pollution contribute to ecological degradation, while also suggesting potential restoration pathways. To validate its applicability, the framework was applied to four major urban streams in Seoul, where indicator-based analysis and correlation mapping revealed meaningful linkages among urban drivers, biodiversity, air quality, and civic engagement. Ultimately, by integrating large-scale text mining with causal inference modeling, this study offers a transferable approach to support adaptive planning and evidence-based decision-making under the uncertainties posed by climate change.

 

 

 

Comment 3: The map would make more sense if showing the urbanised areas, not just river and forest.

Response: We appreciate the reviewer’s suggestion to improve the clarity of the spatial context by displaying urbanized areas in the map. However, we would like to clarify that the study area—Seoul—is the most urbanized region in South Korea, with over 90% of its surface covered by built-up land, such as roads and buildings. This urbanization is visually dominant and makes it technically challenging to add separate layers that meaningfully distinguish "urbanized areas" from non-urban zones without overwhelming the base map.

Given that the streams examined in this study are all embedded within a highly urbanized matrix, we opted to emphasize rivers and adjacent forest patches to clearly highlight the limited but ecologically significant riparian zones. This choice was made to maintain visual clarity and focus on the target ecosystems of interest. We have clarified this rationale in the revised figure caption.

• Before: Not mentioned
• After: " Note: Although not separately visualized, the surrounding areas are predominantly urbanized, with more than 90% of the land surface covered by buildings, roads, and other impervious structures."

Comment 4: The map would make more sense if showing the urbanised areas, not just river and forest.

Response: We thank the reviewer for this suggestion. However, we respectfully chose to retain Table 1 in the main manuscript. The primary aim of this study is to develop a causal framework through text mining, and Table 1 plays a critical role in presenting how unstructured textual data were systematically categorized into meaningful thematic clusters. The table serves not only as a summary of extracted data but also as direct evidence of the structural logic behind the DPSIR-based model development.

Given that one of the methodological contributions of this research lies in demonstrating the interpretability and traceability of text-mined data, we believe that Table 1 should remain in the main text to preserve transparency and reproducibility. Moreover, this table helps readers understand how representative sentences and keyword patterns were aligned with the identified topics and subsequently mapped onto the DPSIR components.

 

We kindly ask for the reviewer’s understanding regarding the need to include this table as part of the core analytical results.

 

Comment 5: The Problem of Descriptive Method in Text Mining Research

⧠ Comment 5-1. “any supporting literature?”, “where?”

⧠ Comment 5-2. “your synthesis must be supported the published studies?”

Response: We understand the reviewer’s concern regarding the descriptive nature of our text mining results and the perceived lack of citations supporting the interpretive synthesis. However, we believe that this issue arises from the inherent nature of text mining as a data-driven method. The interpretive statements in the Results and Discussion sections were derived from a close examination of 1,001 abstracts and over 7,003 individual sentences. These syntheses reflect patterns and thematic relationships that emerged from our own structured analysis, not from a reinterpretation of conclusions drawn in individual studies.

Therefore, we regard these findings as original insights based on our systematic processing of the literature corpus, rather than secondary interpretations of prior authors' conclusions. Accordingly, we did not attach specific citations to every interpretive statement, as they stem from aggregated patterns across the dataset rather than from any single source.

That said, we agree that certain claims—particularly those making broader generalizations—benefit from support by existing literature. In response, we reviewed relevant sections and added targeted citations where appropriate to reinforce key assertions. We thank the reviewer for drawing our attention to this nuance.

Comment 6: This is very confusing especially in the results – I picked discussion vibes in the results and you will see my comments. (e.g., “In conclusion, the modular grouping of these factors…”)

 

Response: We appreciate the reviewer’s observation regarding the presence of interpretive language in the Results section. Upon review, we fully agree that the paragraph beginning with “In conclusion, the modular grouping of these factors…” contained discussion-oriented interpretations better suited for the Discussion section.

To improve the structural clarity of the manuscript, we removed this paragraph from the Results section, as its core content was already included and more appropriately discussed in Section 4.2. This adjustment avoids redundancy and ensures that the Results section remains strictly descriptive and focused on empirical findings.

 

We thank the reviewer for helping us enhance the organization and readability of the manuscript.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript “Application of a DPSIR-Based Causal Framework for Sustainable Urban Riparian Forests: Insights from Text Mining and a Case Study in Seoul” reads well so far, the suggestions for improvement have been implemented. Good luck with the manuscript and the protection of the urban riparian forests you report on!

Page 1, first sentence: Regarding the more global statement, for which you have not yet found a reference - how about this (https://www.mdpi.com/2571-550X/1/3/21), or maybe you can find something in the long reference list there.

Page 3, second paragraph of ‘Methodology’: Is a reference available regarding catchment area sizes?

Page 4, first paragraph: The words ‘PRISMA’ and ‘SciBERT’ must each be followed by a space before the opening bracket.

Page 5, second paragraph: You changed it to “Three reviewers conducted this assessment to minimize selection bias.”, not “three members of the research team reviewed abstracts to minimize bias.” – which is also fine.

Author Response

Reviewer Comment 1:
Page 1, first sentence: Regarding the more global statement, for which you have not yet found a reference - how about this (https://www.mdpi.com/2571-550X/1/3/21), or maybe you can find something in the long reference list there.

Response:
Thank you very much for this thoughtful suggestion. We have reviewed the paper you recommended and found it highly relevant to our introductory statement. Accordingly, we have revised the sentence on Page 1 to better reflect a global perspective on settlement patterns along rivers. 

Reviewer Comment 2:
Page 3, second paragraph of ‘Methodology’: Is a reference available regarding catchment area sizes?

Response:
Thank you for your comment. Initially, we considered using existing literature to determine catchment area sizes and stream lengths. However, we found considerable variation across sources, which made it difficult to ensure consistency. Therefore, as noted in the preceding sentence, we calculated the catchment areas and stream lengths ourselves using spatial analysis tools in ArcGIS Pro, based on official spatial datasets provided by the Seoul Metropolitan Government. This approach allowed us to maintain methodological consistency and ensure the accuracy of the spatial boundaries used in the analysis. We have clarified this point in the manuscript to avoid any ambiguity.

Reviewer Comment 3:
Page 4, first paragraph: The words ‘PRISMA’ and ‘SciBERT’ must each be followed by a space before the opening bracket.

Response:
Thank you for pointing out this formatting issue. We have corrected the spacing errors in the manuscript by adding a space after ‘PRISMA’ and ‘SciBERT’ before the opening brackets. These changes have been implemented on Page 4 to ensure typographical consistency.

Reviewer Comment 4:
Page 5, second paragraph: You changed it to “Three reviewers conducted this assessment to minimize selection bias.”, not “three members of the research team reviewed abstracts to minimize bias.” – which is also fine.

Response:
We appreciate your attention to detail. We confirm that the revised sentence, “Three reviewers conducted this assessment to minimize selection bias,” more accurately describes the process and has been retained in the final version of the manuscript.