Curiously, most of the accounts identified as bots were tweeting from Brooklyn, New York.
Other top locations for likely bot users included Gundungurra Country in south-east New South Wales, Indonesia, Wynyard in Tasmania, and some capital cities.
Human tweets were most likely to have come from Australian capital cities, such as Melbourne, Sydney, Perth and Brisbane.
About 60 per cent of accounts were very likely to be human, while new accounts created during the election campaign were more likely to be bots.
Surprisingly, the saturation of bots active in Australian politics could be double the rate in the US.
As revealed in one of his previous studies, Dr Graham said about 5 per cent of Twitter users were bots during the first US presidential election debate in 2016.
“Using the same model, using the same metrics and forms of analysis that we’ve used previously, what we’ve found is there’s just a whole bunch more, at least bot-like accounts, that were operating on Twitter (in Australia) during that time,” he said.
“So that’s the first big finding that we’re trying to grapple with at the moment – what’s going on here? Why is the figure so high?”
The US research found also that even though there were far fewer bots than people tweeting during the debate, the average bot was 2.5 times more influential than the average human.
That is, they were more successful at getting retweeted.
“Maybe that goes some way to explaining why the (US) election turned out the way that it did, with all of these bots active and kind of flooding the information space with their own opinions, maybe it could have shifted people’s voting behaviour – but that’s a whole other matter,” Dr Graham said.
“If you’re a person who says, I really think Trump should be the president, he’s got good policies, whatever, and nobody retweets it, then it’s kind of like shouting into the forest and nobody responds.”
However, it was not as simple as Twitter bots being clearly pro-Trump or Pro-Clinton.
Some bots appeared to be controlled by Black Lives Matter activists, which might be typically associated with the Democrats, but then argued that African Americans should vote for Donald Trump.
Other bots and human trolls posted negative news stories about civil unrest, gun violence and political mayhem.
In January last year, Twitter admitted more than 50,000 Russia-linked accounts used the platform to post automated material about the 2016 US election.
It is yet to be determined whether bots swayed the way people voted at the ballot box in Australia on May 18.
Dr Graham said he was still examining the data to see what the Australian bots were tweeting about and whether they were partisan and it was still unknown who created them.
“From a national perspective, the working hypothesis could be that if these are indeed bots, then they’re being deployed by interested parties,” he said.
Felicity Caldwell is state political reporter at the Brisbane Times