Geospatial Mapping of Suicide-Related Tweets and Sentiments among Malaysians during the COVID-19 Pandemic

Round 1

Reviewer 1 Report

The study focuses on analysing the usage of the word "suicide" on Twitter among Malaysians in the context of the COVID-19 pandemic.

The topic is interesting and worth investigating. However, several issues should be addressed. The main issues are that the number of tweets is very small (392) and the results of Vader are not evaluated.

1. The introduction can be considered very long. Please consider dividing it into two separate sections, one dedicated to the "Introduction" and one for "Literature Review".

2. A paragraph summarising the structure of the paper should be added at the end of the Introduction section. 

3. The dates in Figure 1 should be written using the standard ISO format. The information in the figure is hard to understand since the "Number of News on Mental Health" seems 0. 

 4. A reference of footnote should be provided for NVivo.

5. In section 2, the paper specifies that "the collection of data is only available from the third (3rd) to the sixth (6th) 273 of August 2021." The paper should more clearly specify the period for which the tweets have been collected.

6. The volume of analyzed tweets, namely 392 tweets, is very low.

7. The results provided by Vader should be evaluated using standard metrics (Precision, Recall and F1-Score).

Author Response

Please find the attached file. Thank you.

 

Author Response File: Author Response.pdf

Reviewer 2 Report

1- In the abstract, it is important to clearly state the research problem that the study aims to address. It could be helpful to revise the abstract to explicitly mention the research problem that the study is attempting to solve. 2- The introduction is too long. It should be concise and highlight the differences between the current study and related work and provide a clear description of the research being conducted. It would be helpful to shorten the introduction and clarify the main objectives of the study. 3- The study's dataset appears to be small, which may limit the scope of conclusions that can be drawn. 4- The authors should double-check the referencing of figures in the text. The figures are not accurately referenced.

 

Author Response

Please fine attached file. Thank you.

 

Author Response File: Author Response.pdf

Reviewer 3 Report

The manuscript clearly describes the context of using the word "suicide" on Twitter 21 among Malaysians during the COVID-19 pandemic. However, the main issue of this work is the need for more novelty in terms of methods used for analyzing the Twitter text. Since time is a crucial parameter that dramatically affects the results why the authors did not consider this criterion. Some deep learning methods, including LSTM, could be helpful in this work.

Author Response

Please find the attached file. Thank you.

 

Author Response File: Author Response.pdf

Reviewer 4 Report

Author Presented the title “suicide’: Twitter Content Among Malaysian During COVID-19”. Here I would like point out few suggestions.

1.      Why author mentioned the word ‘suicide’ in title ?.In abstract, Author did not represented the analysis learning algorithm or processing methods.

2.      What is the final outcome of this model ? is there any numerical or quantity results ? these details should be included in the abstract

3.       All the figures should be cited properly in the manuscript and reference sections.

4.       In Fig.1, Author represented in different language. As a reviewer we can not identify the meaning of x-axis details. So its better to represent in common known language like English

5.      The raw data collected from twitter. Here we cannot find the pre-processing details of the raw data. Before content analysis, author need to apply pre-processing techniques and that details could be added in the manuscript.

6.      In fig.5 author represented GOTD. What is GOTD ? similarly many abbreviation  was missing in the manuscript.

7.      Sentimental analysis of twits information was discussed in section 3.2. But here what are all the algorithms were applied ? atleast few machine learning algorithms can be explained.

8.      In Figure.7 pie chart representing and classify the total outcome of the inputs data. How these values are calculated? What is the performance evaluation of this ?

9.      Author can include the mathematical models for this.

Author Response

Please fine the attached file. Thank you.

 

Author Response File: Author Response.pdf

Reviewer 5 Report

The study focuses to understand the context of the word "suicide" usage on Twitter in Malaysia during the COVID-19 pandemic. The paper is generally well-written but the following must be addressed before going further:

1- Title: The title should be written in a better way. Like, "Geospatial Twitter Sentiment Analysis to Detect Suicidal Tendency among Malaysians during COVID-19."

2- Dataset: How the dataset was filtered from 18000 to just 392? On what basis? What about the rest of the tweets? A proper justification is needed otherwise it can question the significance of the study. 

Why there was a need to convert the tweets into English? While there are NLP approaches available for Non-English languages, such as;

https://doi.org/10.3390/bdcc7010016

  https://doi.org/10.18280/mmep.090617

 

Add a subsection related to the Data Preprocessing phases like cleansing, labeling, etc.

3- There are several NLP approaches, please justify the reason for selecting VADER over the other. 

4- The study reflects the +ive/-ive sentiments towards Govt. bodies and religion, it is recommended to add a statement as a disclaimer so relevant bodies or the people should not be hurt.

5- State the implications of the study clearly.

6- Comparison to the similar type of studies missing. Such as:

https://doi.org/10.3390/bdcc7010016

  https://doi.org/10.18280/mmep.090617

Author Response

Please find the attached file. Thank you

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I would like to thank the authors for the changes made. However, additional details should be provided:

- The authors state that the results of the classification have been evaluated in terms of Precision, Recall and F1 Score. The paper should also mention how many messages have been used to compute the values for the metrics. Additionally, the paper should explain how the messages that have been used for computing the metrics have been manually annotated by human raters and how disagreements have been handled.  

Author Response

Reviewer 1

I would like to thank the authors for the changes made. However, additional details should be provided:

- The authors state that the results of the classification have been evaluated in terms of Precision, Recall and F1 Score. The paper should also mention how many messages have been used to compute the values for the metrics. Additionally, the paper should explain how the messages that have been used for computing the metrics have been manually annotated by human raters and how disagreements have been handled. 

 

Response:

Thank you for your suggestions. We feel that our manuscript is getting better with the comments provided. We are really grateful and blessed to have meticulous reviewers to evaluate our manuscript.

To clarify your comments, we described in details the number of messages and the criteria of human raters and how we handle the disagreement during the validation procedure; between line 314 -126.

 

Author Response File: Author Response.pdf

Reviewer 3 Report

The manuscript can be accepted.

Author Response

Reviewer 3

The manuscript can be accepted.

Response:

Thank you for your all recommendations and support.

 

 

Reviewer 4 Report

Author answered all my questions. 

Author Response

Reviewer 4

Author answered all my questions. 

Response:

Thank you for your all recommendations and support.

 

Round 3

Reviewer 1 Report

I would like to thank the authors for the changes made.