Prior literature has investigated the connection between school shootings and factors of familial trauma and mental health. Specifically, experiences related to parental suicide, physical or sexual abuse, neglect, marital violence, or severe bullying have been associated with a propensity for carrying out a mass shooting. Given prior research has shown common histories among school shooters, it follows that a person's violent tendencies can be revealed by their previous communications with others, thus aiding in predicting an individual's proclivity for school shootings. However, previous literature found no conclusions were drawn from online posts made by the shooters prior to the mass shootings. This thesis applies NVivo-supported thematic analysis and Natural Language Processing (NLP) to study school shootings by comparing the online speech patterns of known school terrorists versus those of non-violent extremists and ordinary teenagers online. Findings indicate that out of all the possible NLP indicators, conversation, HarmVice, negative tone, and conflict are the most suitable school shootings indicators. Ordinary people score eight times higher than known school shooters and online extremists in conversation. Known shooters score more than 14 times higher in HarmVice, than in both ordinary people and online extremists. Known shooters also score higher in negative tone (1.37 times higher than ordinary people and 1.78 times higher than online extremists) and conflict (more than three times higher than ordinary people and 1.8 times higher than online extremists). The implications for domestic violence prediction and prevention can be used to protect citizens inside educational infrastructure by linking the flagged accounts to the schools or colleges that they attend. Further research is needed to determine the severity of emotional coping displayed in online posts, as well as the amount of information and frequency with which weapons and killing are discussed.
| Identifer | oai:union.ndltd.org:ucf.edu/oai:stars.library.ucf.edu:honorstheses-2553 |
| Date | 01 January 2023 |
| Creators | Do, Quan K |
| Publisher | STARS |
| Source Sets | University of Central Florida |
| Language | English |
| Detected Language | English |
| Type | text |
| Format | application/pdf |
| Source | Honors Undergraduate Theses |
Page generated in 0.0024 seconds