Dr Abhay Kumar Singh is a senior lecturer at Macquarie University’s Macquarie Business School.
Many of us tune in to the ABC’s Q&A at 9:30pm Monday to watch a panel of experts discuss current issues relevant to Australian society.
While the socio-economic commentary is always compelling, I’m more interested in the program’s Twitter feed because I’m involved in multidisciplinary research in the area of data science including social media analytics and financial data analytics.
I wanted to see whether text-mining can be used to demonstrate the relationship of the federal and state Covid-19 messaging and the Q&A panel expert thought leadership with people’s responses via Twitter.
(I recently used text-mining in published research: “The Use of Twitter for Innovation in Business Markets to explore both the power of Twitter, using advanced text mining methods, and how innovators use the platform to generate innovative ideas and market their products and ideas.)
Two recent Q&A programs -- Coranavirus: The Cost (March 30) and Coronavirus: Stories from the Frontline (April 6) -- focused on Covid-19.
Considering most of the Australians watch the program on their screens and provide comments on Twitter, it’s important to understand if they are keeping up with the amount of information circulating in the public domain, and their understanding of the pressures faced by healthcare and related sectors.
The program on March 30 attracted comments related to federal government Job Keeper and Job Seeker measures. The subsidy was, as expected, heavily discussed on Twitter. The Basic Universal Income was another heavily discussed topic -- unimaginable three months back.
Stories from the Frontline program on April 6 covered the wellbeing of our medical professionals, the global shortages of PPE (personal protection equipment), and the issues of the general public wearing masks. The Q&A Twitterfeed showed we’re all concerned about the PPE shortages for Australia and there was a high volume of responses to wearing masks in public.
For both programs I analysed a sample of tweets posted during both the programs using word correlation (words used together) such as "stay home", "social distancing", "Ruby Princess”, “community testing”, “flattening curve”, "wash hands", "job keeper", "international students".
Encouragingly, the data suggests that the federal and state government messaging of "staying home" and "social distancing" is cutting through.
But Q&A is just one example of how Covid-19 messaging is reflected in the media in general, and social media specifically.
The rapid spread of Coronavirus (Australia now reaching more than 6400 positive cases of COVID-19 in over 80 days since the first reported case of COVID-19 on 25 January 2020) has resulted in unprecedented interest on social media.
Some interesting observations can be made just be looking at a sample of around 1000 tweets a day containing
the daily trending hashtags in Australia such as #coronavirusaustralia, #coronavirusau, #coronaaustralia, #covid-19au and #covid19au.
Observing tweets using a simple data mining method of n-gram analysis and sentiment analysis over the nine-day period from 31 March 2020 to 8 Apr 2020, it’s evident that Australian public has been on top of the news coverage and also receiving the messages such as “stay at home”, “ washing hands” and “mental health”.
We can also see that some of the issues gaining interest among the public, like the uncertainty with international students and temporary visa holders and of high interest is the fallout of the Ruby Princess cruise and Bondi beach goers.
Another interesting analysis of the tweets is gauging the average sentiment scores based on a simple dictionary method using the NRC Emotion Lexicon. As expected, the scores are quite volatile which is due to the rapidly evolving news coverage and events. On average, the daily positive sentiment over the period was on the decline until we started recording lower number of daily new positive cases. The good news is that the daily average negative sentiment showed decline from 3 Apr 2020.
A different point of view to the sentiment analysis is to analyse the various emotions in the words we use by a breakdown of words used in the sample data along with the emotions.
It’s typical of the “bag of words” based approach to find a lot of positive words and hence positive sentiment but the algorithm could also point out the word usage for other emotions. Of course, the pandemic is a sad topic with the analysis resulting in a high proportion of related emotions including sadness, surprise, disgust, fear, anger and trust.
This research shows how social media and the Twitter in particular are powerful ways in which governments can inform and educate the general public to adopt health measures such as "hand washing" and "social distancing".
Dr Abhay Kumar Singh is a senior lecturer in the Department of Applied Finance at Macquarie University’s Macquarie Business School.
Have something to say on this? Share your views in the comments section below. Or if you have a news story or tip-off, drop us a line at adnews@yaffa.com.au
Sign up to the AdNews newsletter, like us on Facebook or follow us on Twitter for breaking stories and campaigns throughout the day.