OTTAWA -- Out of more than 350,000 comments sent to incumbent candidates on Twitter during the first week of the federal election campaign, more than a quarter were considered toxic, according to a new project from a non-partisan research group.

Sabreena Delhon, executive director at the Samara Centre for Democracy, said the research, which looked only at Twitter, found 20 per cent of the tweets were on the low or middle end of a "toxicity scale," containing insults, sexist language or rude comments.

She said a further seven per cent of the tweets were "severely toxic," including hateful, aggressive comments or threats of violence against candidates or their families.

"If you're on the receiving end of these messages, they are insidious. They are relentless. They are coming at a rate of dozens, hundreds or even thousands a day," she said.

The results of the project so far, which will run throughout the campaign, has found the Liberals are most often on the receiving end of toxic tweets, followed by the Conservatives, the Greens, the Bloc Quebecois and then the New Democrats. More information about those numbers are currently unavailable, but the first detailed report is due later this week.

Delhon said the research found that women seeking re-election as Liberals faced the most toxicity during the period studied and were over five times more likely to receive toxic tweets than men running as candidates for the same party.

The research found Liberal Leader Justin Trudeau is the most targeted party leader with nearly 19,000 tweets aimed at him considered "toxic." Numbers for other party leaders were unavailable.

A spokesperson for Twitter Canada said in a statement the company takes action when it identifies tweets or accounts that violate the company's rules and that it has a civil integrity policy that covers the publishing of misleading content on Twitter.

"We want Twitter to be a place for safe, healthy conversations," the statement said.

The Samara Centre for Democracy partnered with Areto Labs, which has been tracking toxicity online for several years during election campaigns in Canada, the United States and New Zealand.

They developed an artificial intelligence tool that uses machine-learning to review, track and rank toxic tweets received by political party leaders and incumbent candidates during the campaign.

"This means being able to distinguish a rude or insulting tweet from one that is threatening or even sexually explicit," Delhon said.

The two organizations will publish weekly reports on hate online during the campaign.

The Conservatives, NDP and Greens didn't respond to requests for comment on the issue.

"The Liberal Party of Canada condemns all forms of hate and discrimination," spokeswoman Adrienne Vaupshas said in a statement Wednesday. "Hateful and harmful content has no place on social media platforms or in our political discourse."

Delhon said hateful statements and acts discourage voters and candidates from engaging in the democratic process, especially women and members of minority communities.

"That's a pretty big problem because it's hindering equity in representation and limiting the inclusiveness of our democracy," she said.

The Liberal government introduced a new law shortly before the House of Commons rose for the summer that it said would force social media companies to remove hate speech and any statements that are illegal in Canada.

The bill died on the order paper when Parliament dissolved for the Sept. 20 election.

A spokesperson for Facebook Canada wasn't available for official comment, but shared some information and data on the company's efforts to fight hate speech on its platform.

The company's data suggest that it had removed 31.5 million pieces of content for violating Facebook's hate speech policies in the second quarter of this year alone and that nearly 98 per cent of it was removed before users reported it.

The company says it will host virtual candidate training for candidates from all political parties to help them reach and engage with their communities.

Duff Conacher, the co-founder of Democracy Watch, said Canada needs an independent watchdog agency with the necessary resources to enforce the rules that prohibit people and entities from publishing hate speech and false statements about candidates during the election campaign.

"It doesn't do anything to stop them six months later, because the election is over and the negative effect on the election results has already happened," he said.

He said the Liberals' policy of asking social media companies to enforce their own rules is not sufficient enough to tackle this issue that is threatening democracy in Canada.

"The Liberals have left it in the hands of these big social media companies ΓǪ who we know nothing about and who are operating in secret," he said.

This report by The Canadian Press was first published Aug. 25, 2021.

------

This story was produced with the financial assistance of the Facebook and Canadian Press News Fellowship.