Artificial Intelligence and Unsolved Historical Homicides

We have initiated what we hope is a long-term collaboration between University of Alberta (Randy Goebel, Amii/Computing Science) and Edmonton Police Service (Brian Rector), to investigate and develop the application of AI methods to aid in the solution of long-term unsolved homicides (“cold cases”). The initial idea is focused on AI methods for information extraction from historical cold cases. Cold case files are dossiers of a broad collection of data, including paper documents, photographs, DNA evidence, witness reports, records of temporal events, confirmation of relationships and interactions amongst individuals, even video and voice documents.
The initial idea is to use AI methods to create abstractions of the relationships amongst cold case files. For example, methods from Natural Language Processing (NLP) tools can be used to summarize the relationship between all named individuals and all temporal events in a cold case dossier, with the hope that making these relationships explicit can aid the cold case experts in developing new leads to help solve a case.

Faculty Supervisor:

Randy Goebel

Student:

Partner:

Edmonton Police Service

Discipline:

Computer science

Sector:

Public administration

University:

University of Alberta

Program: