A loan of 230 thousand euros was assigned to the "Net Guardian" project, an application developed by the University of Padua together with the Carolina Onlus Foundation, with resources from the Tim Foundation through the Call for Ideas-Research and Education tender.
The project, launched in February and which will end in June 2025, aims to develop an app available to students, school services and associations, based on an Artificial Intelligence model capable of measuring exposure to the risk of online violence in digital conversations.
It involves the identification and classification of risks on all messaging platforms, to activate targeted interventions based on the severity of the verbal aggression detected, evaluating the language used on a scale of 1 to 10, from minimum to maximum risk.
The machine learning algorithm for automated text analysis uses Google's "Bert" deep learning model.
"If the value - explains Gian Piero Turchi, coordinator of the project - fluctuates between 1 and 3 (low risk) the user will be asked to continue inserting subsequent conversations into the App; if the value is between 4 and 7 (medium risk) the user is advised to contact the local services; if the value is higher than 7 the user is notified that he will be contacted by a cyberbullying task force".
Key partner of the project is the Carolina Foundation, dedicated to Carolina Picchio, the first recognized victim of cyberbullying in Italy, committed since 2018 to the digital well-being of the new generations.
Reproduction reserved © Copyright ANSA