Knowledge graphing in forensics: Modeling forensic case trajectories using news media

We employ a knowledge graph (KG) approach to aggregate qualitative information to harness and systematically incorporate contextual information into a forensic investigation. We are interested in tracing the spatial distribution and what happens to unidentified human remains after discovery-relationships and successive movements that are not necessarily captured in formal reports. Our aim is to show that, once it is properly modeled, constructed, and implemented in a scalable graph database infrastructure, a KG can be used to answer a rich set of questions that are of direct interest to researchers, forensic practitioners, and advocates. Our focus here is the development of our KG and methodology. We analyze a dataset of South Texas newspaper articles (n = 370) to search for patterns in the trajectories of migrant case outcomes. We utilize local newspapers because they are a ubiquitous source of information and, especially in the development phase, are not forensically sensitive while, simultaneously, they contain useful contextual information. Our study highlights the United States Border Patrol's jurisdictional purview but also highlights what we term Situational Participants, such as a local who happened upon a decedent. Further, the KG detects consistent gaps in information that can inform where support (financial, personnel) is needed. KGs show great potential to help provide fuller insights into the entities involved and successive actions that characterize forensic investigation in South Texas and may be further extended to other areas of the US-Mexico border and broader humanitarian contexts.