A Software to Extract Criminal Networks from Unstructured Text in Spanish; the Case of Peruvian Criminal Networks

Understanding criminal networks and their activities is relevant for law enforcement authorities (LEAs) to comprehend their complex structures before designing strategies to defeat them. However, getting information to understand these organizations is a difficult matter, which requires a meticulous method to discover and to map the relationships between criminal organizations' members. In this paper, a software to support this process is presented. This software ingests unstructured text written in Spanish, i.e. journalistic articles, and provides a pipeline, which includes Named Entity Recognition (NER), Named Entity Classification (NEC), clustering and co-occurrence to generate a graph. This graph can be analyzed using Social Network Analysis (SNA) getting insights about a criminal organization under investigation. Furthermore, a Peruvian criminal network is analyzed using this software and its outcomes are compared with publicly available results produced by Peruvian LEAs.