December 8, 2020

How Twitter helps (and hinders) bushfire recovery

Researchers study role of social media in rebuilding South Coast communities and supporting survivors

A University of Wollongong (UOW) project to study how social media is affecting the recovery of South Coast communities following recent bushfires has already collected more than 200,000 Tweets for analysis.

The project, #RecoverSouthCoast: Understanding Social Media Use in Bushfire Recovery, is looking at the role of social media in rebuilding bushfire-affected communities, and in the solidarity, community connectedness, and emotional wellbeing of bushfire survivors.

Project leader Dr Robert Ogie from UOW’s SMART Infrastructure Facility is a specialist in the field of disaster informatics and researches on how information and emerging technologies can be used to improve decision making during the prevention, preparation, response and recovery phases of disaster management.

The Global Challenges project employs researchers with expertise in data analytics and machine learning, psychology and mental health, primary health care, social science and linguistics.

“There is so much research looking at social media and how it is used to respond to bushfires as they impact communities,” Dr Ogie said.

“But there is not nearly so much work on how it is used for recovery.”

The first task is to collect relevant Tweets that were posted in the 12 months following the 2019-2020 bushfires on the South Coast, from the Shoalhaven to the Victorian border.

The second task is to sift out the “noise” – Tweets that contain irrelevant or trivial content – before the analysis can begin.

The team is using a number of methods to ensure that the data is relevant.

Since only about 1 per cent of Tweets are geo-located, they will be looking for hashtags using South Coast locations such as #Cobargo, #Shoalhaven or #Bega.

In addition, they will conduct keyword analysis for the relevant content so that – for example – if tourism and bushfires are used together in a single Tweet, that data will be captured.

Once collated, the researchers will look at the sentiment of the data as well as looking for six emotions – anger, disgust, fear, surprise, happiness and sadness.

“We want to see how those change and fluctuate over time,” Dr Ogie said.

“One of the really interesting things is to see how social media can actually complicate recovery.

“To this end, we are developing an algorithm that might detect messages which tend to be troubling for people.

“In the future, we would like to develop ways of responding to these Tweets in real time.”

The origins of the Tweets will be sorted into four categories – government agencies (such as the NSW Rural Fire Brigade), NGOs (such as the Salvation Army or the Red Cross), citizens, and politicians.

“We will see which of these categories contribute positively or negatively to the recovery process,” Dr Ogie said.

“Our machine learning algorithm would automatically classify tweets into different categories of recovery activities such as economic recovery, mental health, reconstruction, just to name a few.

“Through analysis of the results, we can determine what aspects of recovery are most salient to the people and how that has changed over the period under investigation. Interviews with the bushfire-impacted communities will further enrich our findings.

“We are aiming to go to the next phase of the research which will create a real time system to detect and respond to troubling messages.”

Social media platforms were used in the recovery processes by keeping communities connected, Dr Ogie said, helping them share experiences, and to access information and resources for rebuilding communities.

“This includes mobilising donations of goods and money, encouraging tourism, expressing feelings, seeking assistance, and showing empathy or solidarity for those requiring emotional support,” he said.

“That is why it is important to understand how bushfire-affected communities engage with social media content and how this engagement supports community recovery.”

The first stage of #RecoverSouthCoast will run for 12 months.

Other members of the team are social scientist Dr Joshua Whittaker, linguist Dr Alison Moore, SMART researcher Dr Mehrdad Amirghasemi, primary nursing care specialist Dr Sharon James and psychologist Professor Mitch Byrne.