Theft Gang Discovery Using Co-offending Knowledge and SNA

A link mining study on a theft network is done in cooperation with Bursa Police Department in Turkey on more than 100,000 crimes and 6,000 persons. Group Detection Model (GDM) is based on co-occurrences of offenders in police arrest records for generating possible theft networks. Out of thousands of groups detected, only 63 ad-hoc theft groups are selected and introduced to the police experts. To evaluate these findings, one theft network is focused, a preliminary judge decision is obtained for phone tapping and group members phone conversations are eavesdropped for ten weeks. After verification of evidences, Operation Cash is launched. The police arrested 17 people, recovered worth $200,000 of stolen goods, and cash worth $180,000. Conviction and evidence showed that ruling members in offender networks can be detected using GDM. GDM uses a graph generative model and Social Network Analysis (SNA) for link mining in crime data.