Social Media Data in Urban Design and Landscape Research: A Comprehensive Literature Review

Round 1

Reviewer 1 Report

 

The authors aim at presenting “A Bibliometric Literature Review” on Social Media Data in Urban Design and Landscape Research”

As I am no expert in the very subject of the study I’m in the following going to restrict myself to review the bibliometric part of the work.

The title initiates my first criticism: the composite “Bibliometric Literature Review” seems not to make sense for me and does not adequately describe the study. Bibliometrics is the quantitative investigation of a publication corpus and literature review the qualitative exploration of a corpus relevant for a field of research. So the title could mean a literature review informed by a bibliometric study. But the present manuscript cannot be regarded to fall in this category. In the first part until section “Results 3.2” the authors try to present a bibliometric study of the research field indicated in the title. In a second part starting from “Results 3.3” they present a literature review of the main tools that are currently used to analyse social media. But I can find no connection between these parts. So it is not clear why the authors did the bibliometric analysis in the first place. Or maybe the manuscript should be split into two very disparate papers. Already in the “Introduction” the authors put these two aims besides one another on lines 64 to 69 and elaborate on bibliometric methods, but more precisely visualization tools, on lines 70 to 74, and on social media analysis tools on lines 75 to 89.

This said, I restrict myself to the first part. There I find severe methodological weaknesses:

In the section “2. Methods – 2.1. Data collection” the Web of Science search formulation is not given in a clear and reproducible way. Taking the query fragments given I tried the following query formulation in WoS which seemed adequate to me:

(TS= Urban or TS= Landscape or TS=park or TS="public space" or TS="open space") and TS="social media"

This yielded 2785 hits with the suggested reasonable restriction to the document types Article and Review, i. e. more than twice as many as the authors report.

And then it is not clear if the authors report some research areas contained in the dataset or restricted their search to them – why should they do this? But with this restriction I get only 767 results – much less than the authors.

Thus, with this uncertainties there is no real basis for the application of the - nevertheless adequate - bibliometric visualization tools CiteSpace and VOSviewer.

But even it the data basis would be sound there are still weaknesses in the Results section.

3.1 Publication Growth Trend:

Line 141 why is the increasing growth rate foreseeable?

Did the authors use any bibliometric methods or simply inspection of keywords and titles to identify the three temporal stages and to characterize their contents?

3.2 VOSviewer and Citespace Results

3.2.1

Did the authors use author keywords, keywords plus or both?

I wonder why in figure 2 the keyword “social media” is not prominently present, when it has been ingredient of all search queries?!

What is the significance of figure 3 and the list of most-cited keywords?

3.2.2

For non-experts in using CiteSpace it would be important to know how the timeline graph analysis is done, especially how the concepts #0 to #9 are created or retrieved(?), how Fig. 4 should be interpreted apart from simply listing the concepts, and why it especially concerns research after 2017.

3.2.3

The burst keywords citation analysis seems interesting, but the conclusions in lines 204 to 207 are not clearly derived from Figure 5.

3.2.4

The authors do not inform the reader about how CiteSpace is doing the co-citation analysis and what is its significance. In table 1 the list of “top 15 articles” according to “counts” (what does this number signify?) is only given with incomplete bibliographic data, e. g. without publication year, and only two of them (Rozas, Richards), are found in the references and therefore discussed in the following sections. So the reader has to speculate on a possible connection of the now finished bibliometric part and the following literature review.

 

In summary I recommend that the authors do major revisions, i.e. a thorough and reproducible bibliometric study of the chosen subject and make clear how their literature review is guided by the respective results.

Author Response

Thank you very much for your advice, we have revised our research methodology and title based on your suggestions.

1. The title of this study was changed to a comprehensive literature review in which the bibliometric approach appears only as an aid, and we removed the section on VOSviewer, where we felt that Citespace's co-citation analysis, timeline view and burst keywords were sufficient to corroborate our study. Also, we have reordered the article, with a systematic review first, followed by a complementary bibliometric visualization.
2. In the data collection section, we updated the data search, which was a mistake on our part, stemming from the fact that we changed the title name before the peer review, but we did not change the search in that section.
3. We have removed the content of the original 3.1.
4. We have added explanations of timeline view and Burst keywords, and since our systematic review is placed in front of the bibliometric visualization, we also discuss that combined with the systematic review.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments for author File: Comments.pdf

Author Response

Thank you very much for your advice, we have revised our research methodology and title based on your suggestions.

1. The title of this study was changed to a comprehensive literature review in which the bibliometric approach appears only as an aid, and we removed the section on VOSviewer, where we felt that Citespace's co-citation analysis, timeline view and burst keywords were sufficient to corroborate our study. Also, we have reordered the article, with a systematic review first, followed by a complementary bibliometric visualization.
2. Bibliometric software or bibliometric visualisation is a more appropriate alternative name
3. We have added a clear statement of the research objectives in the last paragraph of the Introduction
4. We have added explanations and discussions for co-citation, timeline view and Burst keywords

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper presents bibliometric analysis of urban design and urban landscape research based on social media data using VOSviewer and Citespace. Research was conducted on dataset which consists of 1220 articles and reviews from SSCI, SCIE, and A&HCI based on the Web of Science core collection.

The text is appropriately formatted into logical chapters and paragraphs and is written in English language. When writing the paper, appropriate scientific and professional terminology was used. The researched problem is well described, and the motivation for the research is stated.

The research was conducted using standard bibliometric methods on data collected from Web of Science Core Collection database. Data source is precisely described, and standard bibliometric methods were applied in this research.

Paper also presents levels of implementation of location-based social media, text mining, and image vision in research of urban design and urban landscape.

This is interesting paper which presents analysis of the development of research in scientific literature about urban design and urban landscape from year 2009 to 2022 acquired using standard bibliometric methodology on data collected from Web of Science Core Collection database. Paper presents keywords co-occurrence analysis and display change of research trends in field of urban design and urban landscape.

Minor language proofing is necessary and detected problems are highlighted in yellow in attached document.

Comments for author File: Comments.pdf

Author Response

Thank you very much for your reminder, this article has again been optimized for article presentation

Round 2

Reviewer 1 Report

The authors claim to have revised their research methodology and title based on my suggestions.

This is partly true:
I think the change of title is appropriate, as also the omission of the visualizations using VOSviewer (the abstract has to be changed accordingly)
I think the order of the paper is now  better - presenting the main part - the literature review - first and the bibliometric evaluations using CiteSpace only as supporting information after it. The authors have added some explanations concerning the bibliometric approach including the tool CiteSpace and some illuminating interpreation of its results in the section 4.1. And they added even more to the main part, the literature review.

But my overall impression is that the authors very much hurried to make this revised version and therefore the English style suffered, but even more  
the structure and numbering of the passages, especially on page 5. These errors need to be corrected.
And Table 1 has still incomplete metadata especially when only 5 of the 15 articles  - not "most" as claimed on line 487 - are contained in the list of references.  

But my main criticism concerning the bibliometrics part in the first version of the manuscript is still valid:
The Web of Science search is not reproducible, but still the basis of the later bibliometric evaluations which are therefore still questionable.
I do not know why the authors are not able to present the exact search terms that were the basis for their result set.
Only this can make the bibliometric treatment plausible.

Author Response

Thank you very much for your advice.

1. Originally, I was confused by the first suggestion you mentioned, Page 5 didn't have any typesetting structure and numbering of the passages problems until I downloaded the manuscript.pdfgenerated by the system. The manualscript.docx is the correct version.
2. Regarding the incomplete metadata in Table 1, we have read most of the review articles in urban and landscape research based on Citespace. We believe that the metadata is relatively complete. The reason is that we try to explain the topic and time of the study, while the journal title and other information are not necessary. We have added an explanation of why the literature referenced by the systematic review only has some explanations from Table 1. In the systematic review, our research focuses on NLP and CV deep learning.
3. Regarding the query about abstract and Data Collection, we updated the abstract, and optimized the explanation of that section in the Method section and we made a PDF for you to explain how we search for the Citespace data, which was resolved after the first round of modifications.

Author Response File: Author Response.pdf