The investigation responds to the rise of "fake news", an imported term to talk about false news that has become popular because of its impact and the misinformation that have generated from social networks in various political and social events such as the Brexit referendum or elections Generals in Spain from April 2019.
Researchers Miguel Molina Solana and Juan Gómez Romero, from the Department of Computer Science at the University of Granada (UGR), together with several collaborators from Imperial College London, have presented a study on the use of Artificial Intelligence to detect this false news On twitter.
The work, which has been published by the international magazine «IEEE Access» and reported by the University of Granada in a statement, mathematically analyzes what features Twitter messages contain that contain false information and proposes a computer system to detect lies or hoaxes.
The approach moves away from the systems of journalistic verification, which require a huge work of data verification and review of the newspaper archive. “Analyzing the content of the tweets automatically is very complicated, since it requires studying whether the author is simply being ironic or really trying to make false news true,” adds Molina, who stressed that Artificial Intelligence can help in this homework.
«Experiments have shown that users who distribute false information intentionally behave differently than normal»
The researchers decided to use, together with the content of the tweet, the data available on it and on its author - metadata - to take into account the number of followers at the time of publication, the date of registration in the social network or the number of capital letters and icons used. "Experiments have shown that users who distribute false information intentionally have a different behavior than normal," says Molina, who has detailed that anomalous behavior appears in properties that can be measured as the number of contacts or favorite tweets .
The study notes what features can be used to help detect potentially false news automatically. The work uses Twitter data on the 2016 presidential elections in the United States collected by the authors themselves and the findings have also been presented at the Truth and Trust Online conference organized by London, Twitter, Facebook and the University of Cambridge, among others .
0 Comments