LIVIU: Intelligent mass shooting information extraction from news articles

Summary

Where were you when the last mass shooting occurred? More than 1200 U.S. citizens from 1966 until today were found in the wrong place at the wrong time, not suspecting the upcoming mass shooting they would be part of. No one would expect that on 10/24/2014, Jaylen Fryberg, 15 years old, would shoot and kill five students at Marysville High School, including two of his cousins and three of his friends, before committing suicide. Unfortunate victims, families loosing beloved members, survivors traumatized for life. Could a mass murder be avoided? Are there measures to be taken to prevent the next tragic incident? Aiming towards understanding these events, the need for quality data is clear and urgent and constitutes the reason behind this The- sis, which aims at providing a tool that generates a self-maintained, automatically created dataset, by collecting mass shooting related information given online news articles as input. The LIVIU system we propose is an end-to-end data retrieval and extraction system that, given links to online articles, manages to scrape their contents, performs linguistic analysis using machine learning techniques paired with hand crafted rules, and eventually generates a data collection with key information concerning the attacks, namely the name of the shooter, the location, the date, the number of the victims, and the weapon used to carry the unfortunate event. Our experimental results show that the data collection generated by the LIVIU system significantly resembles the ground truth dataset (to which is compared) and can confidently be used as the first step of an investigation project.