Background: In the last two decades, human trafficking has increasingly taken place on digital infrastructure, with recruitment, advertisement, coordination, and payment increasingly being done by way of social media, encrypted messaging apps, open-web classified sites, and dark-web forums. The field of investigative practice has adapted to these two lines of analysis, namely social network analysis (SNA) and dark-web/open-source digital surveillance, both of which have an evidence base that has yet to be synthesized together with a set of research questions.
Objective: This study aims to identify potential trafficking victims, trace the evolution of trafficking patterns, disrupt organized networks of trafficking and exploitation using network analytics and dark-web surveillance, and describe the ethical and privacy protections needed for responsible use of these tools.
Methods: A qualitative, document-based research design was used. Systematic searching and screening of peer-reviewed journal articles, preprints, and institutional reports (United Nations Office on Drugs and Crime, International Labour Organisation, European Commission, Eurostat) and reputable practitioner sources (Chainalysis, Elliptic, Polaris Project) were conducted, followed by an analysis of the sources in relation to five pre-specified research questions. The primary surveys, interviews and original network datasets were not collected, and the unit of analysis was the published study or statistical release.
Results: It was found that certain structurally important actors whose removal would provide disproportionately disruptive value to the network could be repeatedly identified using centrality-based network methods and selectively enforced. Construct validity and label reliability were questioned, but increasingly powerful machine-learning classifiers were based on online advertisement textual, visual, and stylometric features paired with their relational features. Blockchain analytics introduced a brand-new evidentiary channel – the cryptocurrency ledger is not only permanent but also public, even if the wallet owner is not. Platform-policy interventions that tried to curb trafficking advertising, primarily the closure of Backpage and the passage of FOSTA-SESTA, have had mixed and at times counterproductive results, both in terms of victim identification and in terms of providing safety to the general population who engage in sex work. All the analytic gains found were paired with proven risk of algorithmic bias, privacy violations, and misidentification.
Conclusion: Summary: Network analytics and dark-web surveillance are more properly viewed not as automated systems of surveillance, but as intelligence-augmentation systems that must be monitored, authorized, and protected, survivor-by-survivor. This study aims to address the following 5 research questions sequentially in Sections 4 through 9.