Gang Ontology Project

The Gang Otology Project focuses on extracting information from images posted on social media by gang involved youth in order to find out where help or other forms of intervention might be necessary.

To this aim we create a list of visual concepts that are related to gang life and either indicate violence or provide relevant background information. We then train machine learning models to automatically extract these concepts from images and perform a qualitative analysis to better understand the role of these concepts in online gang communication known as internet or cyberbanging. Finally, we plan to augment our previous work on aggression detection for social media posts that have images by considering the visual contents in addition to the posted text.