Software definitely helped cause the pernicious problem of fake news – but could also ameliorate it, says data expert Emil Eifrem, in a guest blogpost.
Time was it was the tabloid newspapers that cornered the market when it came to fake news (remember ‘Freddy Starr Ate My Hamster’?). These days it’s a whole industry in itself – with the technology in the hands of people who don’t even pretend to be objective journalists.
In fact, traditional journalism’s decline has meant the loss of checks and balances, allowing fake news a free and disruptive reign. Technology advances also supports the ways misinformation is spawned and travels so rapidly, as social media is after all about active sharing. According to one of the biggest studies to date on the problem, conducted by researchers at MIT and published in March 2018 in the journal Science, the truth takes six times longer to be seen on Twitter than misinformation.
Researchers tell us lies are 70% more likely to be retweeted than the truth — even when they controlled for factors such as whether the account was verified or not, the number of followers, and how old the account was. This is not good news for society, democracy and human knowledge, one can argue.
Interestingly, while technology certainly is an enabler of fake news, it may also be the answer to helping combat it. Specifically, graph database technology, a powerful way of recognising and leveraging connections in large amounts of data, may offer some hope of salvation.
Indeed graph software is already used by legitimate investigative journalists: it helped the International Consortium of Investigative Journalists track its way through the TBs of data known as the Panama and the Paradise Papers, for instance. But graph software also turns out to be a way to potentially combat fake news.
Visualising patterns that indicate fake content
It’s reported that Russia used social media in a bid to influence the 2016 US presidential election – and with graph technology’s help, the US’s NBC News has uncovered the mechanism of how that was achieved.
That’s because NBC’s researchers found that the key to detecting fake news is connections – between accounts, posts, flags and websites. By visualising those connections as a graph, we can understand patterns that indicate fake content. The group behind the Russian trolling was small, but very effective, working to leveraging Twitter with popular hashtags and posting reply Tweets to popular accounts to gain traction and followers. In one account, for example, of 9,000 Tweets sent, only 21 were actually original, and overall 75% of all the material were re-tweets, specifically designed to broadcast the messages to as wide an audience as possible. While some accounts posed as real-world citizens, others took on the guise of local media outlets and political parties.
When graph software was used to analyse the retweet network, it revealed three distinct groups – one tweeting mainly about right-wing politics, a second group with left leanings, and a final group covered topics in the Black Lives Matter movement, in an invidious but effective triangulation of extreme content sharing and emotional manipulation.
At internet scale, fake news is just too hard to spot without the right tools. We should look to graph technology, specifically designed to expose connections in data, as a possible way of helping to address the issue.
The author is co-founder and CEO of Neo4j, the world’s leading graph database (http://neo4j.com/)