A new AI-based tool named LOLA has been shown to rapidly root out cyber-bullying and hate speech online, and may even help detect fake news.
One of the main problems with the spread of fake news and harassment online is that compelling stories can spread rapidly via social media if it goes viral. Harmful content can reach millions of people before it's taken down, spreading misinformation or harassing messages that can cause very real harm to people. This has become particularly dangerous during the Covid-19 outbreak, as medical misinformation has been spreading via social media.
A new digital tool developed by the University of Exeter Business School may help to solve this problem by automating the detection of fake news and harassment. The system named LOLA uses AI to classify the emotional context of the language used, and claims a 98% accuracy with detecting harmful behaviour such as cyber-bullying and overt racism.
The tool was developed by a team led by Dr David Lopez from the Initiative for Digital Economy Exeter (INDEX) in response to the fact that the increasing volume of information on platforms like Facebook and Twitter are making it harder to detect abusive behaviour.
Dr Lopez commented: "We believe solutions to address online harms will combine human agency with AI-powered technologies that would greatly expand the ability to monitor and police the digital world. Our solution relies on the combination of recent advances in natural language processing to train an engine capable of extracting a set of emotions from human conversations (tweets) and behavioural theory to infer online harms arising from these conversations."
LOLA has recently also been used to help weed out fake news about Covid-19 by detecting the tones of fear and anger in posts about the outbreak. It produced a list of potentially fake tweets about Covid-19, and ranked them on a scale based on the level of toxicity, obscenity and insult used.
Dr Lopez hopes that LOLA can be used by digital companies to discover harmful messaging before it spreads online and stop it before it causes harm. "The ability to compute negative emotions (toxicity, insult, obscenity, threat, identity hatred) in near real time at scale enables digital companies to profile online harm and act pre-emptively before it spreads and causes further damage."
Source: Written based on press release
