The next session at AoIR 2019 starts with our paper on Twitter activity patterns in the 2019 Australian federal election, and I presented the first part of this so I didn’t blog it, but the slides are below.
My colleague Dan Angus has now taken over, and he presents his insights into the major topics being discussed in the tweet data. These divide into various policy topics that are both supportive and critical of the current government, and discussions about the electoral process; such themes and topics are of course unevenly distributed across the different parties and the campaign period, but there was no single major defining issue throughout the election.
Further, my colleague Tim Graham also drew on the Botometer tool to investigate the presence of bots in the dataset. Some 9% of accounts we encountered were rated as 0.6 or above by Botometer, and many such suspected bot accounts were created very recently before the election – this might mean that Botometer simply flags never accounts as bots, or that there were actually bot accounts created specifically to engage in the election.
Such bot accounts tended to be polarised in the topics they engaged with: there were far less likely to be critical of government-aligned topics, and much more likely to be supportive of its pro-coal policies. But this needs to be further investigated before we are comfortable pointing fingers at any one actor.
