Inductive and transductive link prediction for criminal network analysis

Zahra Ahmadi, Hoang H. Nguyen, Zijian Zhang, Dmytro Bozhkov, Daniel Kudenko, Maria Jofre, Francesco Calderoni, Noa Cohen, Yosef Solewicz

Abstract: Highlights•We study the problem of crime linkage and co-offender prediction from two perspectives: transductive and inductive.•We compare the effect of different similarity metrics in unsupervised transductive link prediction.•We propose an inductive and transductive link prediction framework based on graph neural networks and node attributes.•We share an open-source network analysis tool for network analysis, specifically helpful in criminology.•We share a huge anonymized dataset for a burglary case.

External IDs:doi:10.1016/j.jocs.2023.102063