Propaganda bots are something we’ve come to be very aware of in South Africa in the past year, thanks to the Gupta family and its links. Telling the difference between real and fake can be difficult, which is why anew propaganda bot checking platform has been introduced.

Botcheck.me uses machine-learning tech to pick up on the tweeting activity of certain accounts to determine whether they have traits of propaganda bots or not.

“Inherently, the difficulty of identifying bots on a social media platform like Twitter is the fact there is no way of fully knowing what a bot looks like. Unlike academic datasets, there is no ground truth or labels for these accounts. This is a chicken-and-egg problem: you can’t identify bots if you don’t know what bots look like, and you don’t know what bots look like if you can’t identify bots,” Botcheck.me creators, Robhat Labs, describe in a blog about how its analysis works.

“In order to get around this dilemma, we identified accounts with certain suspicious behavior as high-confidence bot accounts. Behavior such as tweeting every few minutes in a full day, endorsing polarizing political propaganda (including fake news), obtaining a large follower account in a relatively small time span, and constant retweeting/promoting other high-confidence bot accounts are all traits that lead to high-confidence bot accounts. These are the accounts that we aim to classify and bring to the attention of the Twitter community.”

Unsurprisingly, most of the content linked to US-based bots are related to President Donald Trump, including the Make America Great Again (Maga) hashtag and his personal Twitter account.

Trump and his supporters have been known to spread false information and claims publicly during the run up to the US elections last year and since Trump’s inauguration in January.

Botcheck.me is available as a Chrome extension.

[Image – cc 4.0 Discott]