University of São Paulo: Digital platform helps detect fake news
A virtual platform developed by researchers at the Center for Mathematical Sciences Applied to Industry (CeMEAI), at the Institute of Mathematics and Computer Sciences (ICMC) at USP in São Carlos, helps the population to detect fake news with 96% accuracy. Marcos Jardel Henriques, a PhD student in Statistics at the Institute of Mathematical Sciences and Computing (ICMC) at USP in São Carlos, explains the main features of the fake news detection platform to Jornal da USP no Ar 1st Edition.
According to Jardel, the platform was initially designed in light of the current globalized context in which the population is bombarded by information in all media. “We had the idea of setting up a platform for when people had doubts about the veracity of some news, they could copy part or all of the information to verify”, explains Jardel, who recalls the use of statistical models and databases used in the construction of the tool.
He explains that more than 100,000 news articles were consulted and collected to train statistical models with the type of semantics and, mainly, the type of vocabulary used over the last few years. “Then they were tested in other databases prepared by other research groups that work with natural language processing”, he says.
To use the platform, he suggests using as much information as possible to be analyzed. “The more information you put in, the better the platform will understand the context, the semantics, whether the text is well prepared or made in a simplistic way”, he explains. It also highlights that the probabilities demonstrated by the tool are 96% accurate. “It’s good to remember that, as the months go by, we have to recalibrate and update the platform because new information and new names emerge”, he adds.
Cryptocurrency market and covid-19 monitoring
The project was developed during the Professional Master’s Degree in Mathematics, Statistics and Computing Applied to Industry (Mecai) and two other platforms were also developed: one for anticipating the market value of cryptocurrencies and another for monitoring of deaths by covid-19 at the municipal level. The news analysis tool was developed by Jardel and also has the direction of the ICMC professor, Francisco Louzada Neto, and also the doctoral student Daniel Camilo Fuentes Gusmán.