Finding Bad Guys on the High Seas with Data Fusion

Disparate data about vessel behavior, sanctions and ownership come together with Entity Resolved Knowledge Graphs to reveal otherwise imperceptible insights in maritime data.

Which vessels are transacting with whom, where and when?

To identify illegal fishing and sanctions-busting activities like transloading oil, analysts need to be able to assemble and combine disparate maritime data, often intentionally obfuscated, to reveal hidden connections and surface actionable intelligence.

Enter the Entity Resolved Knowledge Graph (ERKG). In this case study, the ERKG acts as a force-multiplier, not only bringing analyst attention to sanctioned entities, but also revealing networks of collaborating vessels and entities.

THE ARCHITECTURE

ESRI data fusion with Senzing entity resolution

Senzing® Entity Resolution: Who Is Hiding In the Data?

Resolving the entity data with Senzing entity resolution before adding it to ArcGIS Knowledge reveals exactly how many entities there are, identifies relationships between entities, and yields important details about activities taking place at specific locations.

Entity Resolution provides the ability to recognize relationships between entities.

Esri ArcGIS Knowledge: Bringing It All Together

Spatial and temporal analysis of Spire Maritime AIS tracking data service identified certain types of Motion Events such as a vessel loitering in one place over time, traveling with or near another ship, or meeting with another vessel. These Motion Events were loaded into Esri ArcGIS Knowledge and linked to the Vessels to populate the knowledge graph with spatial relationships.

Linking Motion Events that took place in a Free Trade Zone (Fig. 1) with all vessels that participated in those events, other events they participated in, and other vessels that participated in those events made it possible to graph the network of interactions with vessels that behaved anomalously in the area of interest.

By combining spatial data and entity resolved financial (ownership) and sanctions data in a knowledge graph, broader patterns of behavior become discoverable through their relationships. The ERKG revealed that previously unremarkable / unknown entities are closely related to known sanctioned vessels and organizations.

Fig. 3 shows the Fu Yuan Yu fleet of Vessels with the other ships they met with, dwelled near, or traveled with, along with their corporate ownership.

The resulting knowledge graph brings to light that the ownership of these other vessels warrants additional scrutiny for potential nexus with the sanctioned entity, Fu Yuan Yu.

Vessel Motion Events on the South China Sea

Fig. 1: Vessel Motion Events on the South

Linking Motion Events to vessels reveals a network of interactions
Linking Motion Events to vessels reveals a network of interactions

Linking Motion Events to vessels reveals a network of interactions

Without entity resolution, this clarity would not be attainable. Many ships would incorrectly appear to be owned by many different companies, and some individual ships would appear to be multiple vessels.

Fig. 4: Spatial data combined with entity-resolved sanctions and ownership data uncovers relationships between unsanctioned and sanctioned entities, highlighting previously unknown vessels that warrant closer scrutiny.