Analyzing and Leveraging Social Media Disaster Communication of Natural Hazards: Community Sentiment and Messaging Regarding the Australian 2019/20 Bushfires

Round 1

Reviewer 1 Report

This study analyzed disaster management through communicative approach, i.e., by using social media and tracing sentiment about Bush fires in Australia.

The study has provided a substantial broad theoretical and conceptual framework for this work.

However, there are a few aspects of the study that need further clarification.

In the sampling, keyword “bushfires” was used for the phase of the disaster. However, for control sample (i.e., before the disaster, was the same keyword used? If so, what are some conceptual expectations for this word to provide the sentiment before the disaster? Similarly, geographic location was delimited to gather tweets. It is known that typically only a small portion of tweets contain geolocation. As a result, authors either had a very stringent approach to their data collection or they bypassed it. Please clarify how data geography has been narrowed down.

Similarly, at times, studies use hashtags for their sampling. What was the presence of hashtag #bushfires in your sample and how was it treated? Similarly, what other hashtags were used and if so, were they included in sentiment analysis processing of a message? If so, how do they “skew” or shape results of sentiment analysis?

For results, authors should include examples of positive and negative sentiment tweets. Similarly, authors should include examples of messages of different stakeholders: i.e., regular users or politicians or journalists who were more likely “drive” the negative sentiment.

For theme analysis, authors should provide tweet examples for each themes.

Authors provide speculative assessment when it comes to rumors, yet no evidence is provided. Authors should run tweet account analysis to establish the percentage of non-human actors (i.e., bots in their dataset). See e.g.: https://botometer.osome.iu.edu/api Otherwise, authors should rethink how they present results on misinformation. Since no evidence is provided.

In discussion authors claim “This suggests that the community was using social media to heighten awareness of issues to affect a government response and boost funding and support for recovery efforts.” Yet their most central figures are media outlets and politicians. Those were the ones who were driving the discussion. Authors should discuss how elites perpetuate and dominate discourses online.

All in all, sentiment analysis is the weakest part of the study—i.e., it is not clear how it helps to contribute to the understanding of the phenomenon—what is the theoretical meaningfulness of this analysis?

Authors provide promising solutions for the future on how to create real-time tools for disaster management. It would be useful to engage in a broader discussion of such tools that combine IoT paradigms and social network analysis, as e.g., in :

Zelenkauskaite, A., Bessis, N., Sotiriadis, S., & Asimakopoulou, E. (2012, September). Interconnectedness of complex systems of internet of things through social network analysis for disaster management. In 2012 Fourth International Conference on Intelligent Networking and Collaborative Systems (pp. 503-508). IEEE.

Author Response

Many thanks for all your comments and suggestions.

We have tried to address all of them.

Author Response File: Author Response.pdf

Reviewer 2 Report

Greetings,    Thank you for providing the opportunity to review this manuscript. I recommend revise and resubmit. Although the work presented is not novel or new and does not, currently, showcase a contribution to the field, there is potential.        Abstract   This research presents a new model to analyze social media discourse through examining Twitter posts during and following Australia’s severe Black Summer bushfires in 2019/2020 from a community on the Gold Coast that experienced the first bushfire event that season. The study proposes that messaging volume is a proxy for the importance of the bushfires and found increases during the bushfire event but declines sharply following the event. However, there was a consistent negative message sentiment. The need for better planning and the links between bushfires and climate change were key message themes. Politicians, broadcast media, and public commentators were central influencers on social media messaging, and this demonstrates the potential of social media to inform disaster response and recovery behavior related to natural hazards.   Questions/Aspects I want answered after reading the abstract:   

1. What is this new model?
2. Is the methodology for Twitter analysis detailed out?
3. Is the context provided for the bushfires and the community being studied?
4. An increase in twitter messages during a crisis is not a novel finding. What is novel about this study? How does it contribute to the fields of crisis communication/risk communication? 
5. How were messages deemed negative? 
6. What is the overall goal of this project? The abstract gives a superficial overview.

  Introduction   Natural hazards- hazard is the event. Catastrophe is the level of impact.    The area of crisis communication, or emergency communication, is more developed than promoting only specific media sources. Expand your literature review to correct this statement.    Why is a numeric citation in the place where the actual author(s) should be spelled out?   How is this research going to build upon the existing literature?    Social Media and Disaster Communication   Correct to social media can shape…you cannot say it does definitively.    Introduce the focus on Twitter within the social media arena. Social media encompasses a number of avenues.    I challenge that Twitter is particularly suitable. Twitter has its qualitative limitations.    Choose another term than infancy. The number of studies related to social media and crises across the globe has increased dramatically over the past decade.    I am curious as to how the author(s) will connect the presence of a Tweet to actions taken by the community.    RQ1 is easy to answer as numerous studies confirm communication activity increases during a disaster.    The research questions do not provide a connection from Tweets to community actions. They focus on the community’s perception and whose Tweet was the most popular.    The author(s) seem to be detailing out Social Network Analysis. This is not a new model.    Methodology   Unsure the significance of including the photos titled Figure 2. How does this assist reader comprehension?   What other terms were utilized to capture the tweets? Was it just bushfire? If so, what if community members simply referred to the crisis as wildfires or fires?   Figure 4 is low quality. Increase the resolution. Also, Twitter’s icon is copyrighted. Make sure you can use it.    Unsure why the author(s) did not utilize Social Network Analysis since they detail out components of the analysis.    The methodology utilized is not new nor is the model. Clarify the contribution.    Results   Make sure all figures are of high resolution.   No examples of how the sentiments were represented via Tweets? Telling me there are themes is too superficial. Showcase exemplars.    The findings, as they currently stand, are not new or novel. If you believe it is then I need more justification as to why and this involves including the literature and how their models did not include the components that yours did.    Also, where are the limitations of the study? Social media is not without its downsides especially during a crisis. Not every community member will have access to a phone or computer. How is this study still relevant despite access issues? 

Author Response

Thank you for your comments and suggestions. We have tried to address all of them.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Great work updating the manuscript.