Presentation POWERED BY Algebra
One of the most basic information we search for concerning a piece of food we’re about to eat involves its origin. We rightly connect the notion of origin to the notion of quality: if the food producer doesn’t want to declare the origin of the food, we immediately suspect there’s something wrong with this produce, and even if we’re not health-obsessed fitness freaks, we don’t want to put suspicious food into our organism. We fear, and rightly so, the possibility of food poisoning.
But what about a different kind of poisoning? As ordinary citizens, we approach things that go into our mouth much more critically than we approach things that go into our head. But much like getting stomach poisoning, we can easily get life-threatening poisoning from suspicious “food for thought”. Just think of people who in the early days of COVID-19 read online that drinking bleach can help fight the deadly disease. And there are so many other examples. Especially in the periods of crises and high emotional engagement, such as COVID-19 or the war in Ukraine now, we pretty much devour everything we can get our eyes on. And while we try to separate the good (trustworthy) from the bad (untrustworthy) content, the process of differentiating the two is highly demanding and given the overabundance of digital sources, often quite unmanageable. This would be considerably easier if, much like with food, we’d be able to tell the origin of the content we read with the same ease we ascertain the origin of the food we eat.
It might seem that in the vastness of digital media space, tracking the origin of information might be impossible, and it is so, for an individual citizen. But it can be managed, and it can be analysed. In this presentation Leo Mršić, PhD (Director AlgebraLAB) and Maja Brkljačić, PhD (Business Development Manager, AlgebraLAB) will show how by using natural language processing, we can deploy the methods of machine learning, in particular artificial neural networks, to automatically or half-automatically detect fake news and their origin, to ascertain their relevance and popularity, as well as to suggest the ways to counter them using the same or similar relevant communication channels.