Kirby Institute Seminar Series presents
Dr George Karystianis
Dr George Karystianis is a research fellow at the Justice Health Program in the Kirby Institute. He obtained his PhD in text mining and epidemiology from the University of Manchester, UK in 2014. Due to a cross-disciplinary skillset in medicine, computer science and epidemiology, he has been designing and implementing large-scale text mining systems for over 10 years in various biomedical fields such as identification of research bias in animal based experiments, key characteristics from epidemiological studies and structured dosage information from medication prescriptions. His current research interests are criminology, mental illness, domestic violence and contagious diseases. He frequently leads teams of experts in international prestigious natural language processing competitions.
Vast numbers of domestic violence (DV) events are attended by the New South Wales Police Force each year and recorded as unstructured free text in the COPS (Computerised Operational Policing System) database containing a wealth of untapped data for DV reporting and surveillance purposes. A text mining methodology was designed, evaluated and implemented to extract mentions of mental illnesses for victims and persons of interest, abuse types, threats and victim injuries from a large-scale corpus of over 492,000 DV police records. The exciting results suggest that text mining can automatically extract targeted information from police-recorded DV events which can be used as inputs in DV predictive models that can assist in the shaping of early prevention and intervention initiatives.