Machine Learning battling fake news in social media platforms

It is imperative to use only verified sources during times where ‘sources claim’ seem to be the norm.

“Want to kill the coronavirus? Get drunk.” This was one of the earliest rumours (among many) going around the block regarding how to combat the COVID-19 threat. The veracity of the statement was nil. The power of sensationalism it carried, however, was not.

It is well-known that alcohol inhibits resistance-building and leads to weakened immunity. Focused on baiting clicks and generating advertising revenue, several outlets have been crossing limits by circulating fabricated and puffed-up headlines in the wake of the coronavirus outbreak. While the culprits made money, several deaths were reported due to consumption of industrial alcohol as people attempted to disinfect themselves and guard their bodies against the virus – trusting the dangerous advice to be a proven scientific fact.

Uncertainty combined with fear leads to anxiety – and panic. It is essential during a global pandemic to strictly regulate misinformation and prevent its spread. Some major internet and social media companies have been making efforts to filter the spread of such fake news, acting as gatekeepers of the information passing through their websites.

During the first month post its emergence, the coronavirus was mentioned in over 15 million tweets – all of which eventually started undergoing rigorous fact-checking from Twitter using machine learning algorithms. Only reliable and authoritative organisations such as the WHO or national health agencies were to be made the primary source of information regarding the outbreak. Conspiracy websites and other news agents often flagged as propagators of ‘fake news’ – with posts that could cause harm or spread panic – had their posts removed. The fact-checking algorithm also detected ‘alternative’ cures and treatment recommendations that fuel widely-believed falsehoods, such as cure by alcohol, consuming untested herbal concoctions or ‘facts’ regarding children’s apparent immunity to the disease.

Facebook, the world’s largest social media platform, too has produced algorithms to protect users from misinformation and other exploitative practices. Advertorial claims, such as homeopathic remedies to cure the virus, were searched for and removed. Bans were also placed on the sale of goods in short supply. Unscrupulous advertisers, who claimed several other unnecessary goods (along with essentials such as masks and sanitizers) would possibly become difficult to acquire owing to the outbreak, were tracked and banned. Instant-messaging service WhatsApp launched a new beta feature in March that lets users search the web for the authenticity of certain messages which might be spurious – thus separating fact from fiction. Facebook-owned Instagram too started identifying and tracking hashtags that were used in posts spreading fake news.

Online message boards such as Reddit, which allows for the generation of ‘conspiracy’ talk, is often the source of several unsubstantiated rumours grounded in racism or other political biases. Although such contents are still available in plenty, several machine learning algorithms are now being used to scour through posts and generate warnings for content that may be unverifiable, false or unsafe.

Several governments as well as health agencies worldwide are partnering with tech giants to limit the spread of opportunistic fake news. The National Health Service in the UK, for example, has tied up with Twitter to accelerate verification processes for over 800 accounts – thereby setting up trustworthy ‘blue-ticked’ news sources to allow users greater clarity. The Indian government, alongside organisations such as the WHO, have also launched WhatsApp chatbots to better educate the masses and deliver authentic information.