Treffer: Identifying Psychosocial Attributes Indicative of Violent Behavior in Students using Deep Learning.

According to the Center for Homeland Defense and Security, in the first half of 2022, 2 active school shooter and 151 non-active shooter events resulted in 150 victims; the previous year's statistic the highest it has been since 1970. Most students displayed signs of mental illness and troubled behavior that was often overlooked. This research seeks to identify signs of a threat in order to distinguish and assist students who are at risk for violent behavior. 30 randomly selected shooters were analyzed using natural language processing with keyword extraction of news reports to identify 28 recurring psychosocial attributes described by the Federal Bureau of Investigation through a WordCloud generator. A feed forward neural network (FFNN), built on Spyder using Python, then uses these traits to recognize and categorize potential growing threats in a student body. Data is collected through deep learning graphological parameters in students' handwriting using a 2D convolutional neural network (CNN), created on Jupyter Notebook using Python. Compared to predictive models such as the Five-Factor Analysis model with an accuracy of 80.5%, the School Threat Assessment System (STAS) uses two simple neural networks to generate an accessible report that quickly identifies students in need of immediate support with an overall accuracy of 97%. STAS is available online to school systems working to increase the safety of their students from within. [ABSTRACT FROM AUTHOR]

Copyright of International Journal of Computing & Digital Systems is the property of University of Bahrain, Deanship of Graduate Studies & Scientific Research and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)