BACK TO WINNERS

Winners For Partner Innovation

First Place – Anti-Mafia Analytics Platform

Winning Team from

$10,000 for Charity of Choice

Play Video

Summary

Tackling organized crime, one node at a time.
Git Hub
Inspiration
Organised crime negatively impacts society from issues such as drug trafficking to corruption to murder. Tackling these issues relies on understanding the complex interconnected web of crime incidents, criminals, and financial activity. Our project seeks to provide an analytics platform that can detect “moles” within the police force and spot fraud transactions. It is clear to see that the identification and subsequent disruption of these criminal organisations would greatly improve society both socially and financially.
What it does
Our analytics platform can see connections between Police Officers involved in an unusually high number of unsuccessful raids and wire tapped phone calls to members of the Mafia. This offers key insights into identifying “moles” within the police force. Furthermore, anomalous payment transaction values within the police force can be used to detect bribery. Wire tap connections can detect who the crucial members of the Mafia are and who to target for maximum impact. Our graph database can drastically reduce the time taken for the structure of a crime organisation to be uncovered, and hence lead to faster and more plentiful abolition of such organisations.
How we built it

We created the schema in Lucid Chart and designated primary keys, foreign keys, and attributes. Since there were no publicly available datasets on the Italian Mafia, we generated representative synthetic data. We performed research into Mafia family hierarchy, main crime activities and geography to make sure our data generated accurately. Using the Python language and associated Pandas library, we created CSV’s for Mafia Members, Police Members, Public Individuals, Financial Transactions, Crime Incidents, Raids and others.

 

The CSVs were loaded into TigerGraph. The schema was created and the CSV files were mapped to nodes and edges.

 

GSQL queries were run to reveal insights of our synthetic data.