Entity Resolution as Mission Critical for Government and Security

The Foundation: Reference Data as Anchor

“When a lot of people think about national security, they don’t think of the rating agency of Moody’s,” Koczan opens. Yet Moody’s Orbis file serves as a backbone for global corporate identity, tying together more than 600 million companies, local tax IDs, and legal entities. That ownership graph goes down to 0.01% stakes and includes directors and managers, which becomes invaluable when agencies need to understand who really controls a company, who sits on its boards, and how those people connect across jurisdictions and sectors.

With a stable baseline in place, analysts could see behavioral shifts in ownership patterns. Circular ownership structures appeared to obscure beneficial owners. Shell companies suddenly showed 90-year-olds or children as nominal owners of multibillion-dollar entities. These anomalies only stood out when analysts had a consistent reference graph to compare against.

The problem is scale. With roughly 5 to 6 million ownership changes every month in the Orbis file, reference data alone cannot keep up with fast-moving threats. This is why dynamic intelligence feeds that add real-time behavior on top of the static ownership graph are an essential part of the solution.

The Dynamic Layer: Real-Time Intelligence

Kesecker describes Kpler’s high-velocity vessel tracking using Automatic Identification System (AIS) data, commodity flows, and crypto movements. “When the government takes an action – whether that’s economic, kinetic, or cyber – that action has a ripple effect,” he explains. The challenge becomes connecting these fast-moving signals to stable identifiers. For example, tracking ships that turn off AIS systems to obscure activity, then linking back to People’s Liberation Army (PLA)-owned or Russian oligarch-owned vessels through ownership data.

On its own, that telemetry is noisy. Its value becomes apparent when you resolve it back to stable entities. A vessel that regularly disables AIS near a sanctioned region becomes far more interesting when you can tie it back to a PLA or Russian operator through beneficial ownership records. The same pattern applies to de minimis shipments and smaller commodities. The signals matter most when they are fused to a trusted backbone of who owns what, where they are, and how they relate to one another.

Making Sense of the Mess

Mike Colesanti distills the integration challenge. “The government rarely has a big data problem. We just have a very messy data problem.” The mission rarely fails because of a lack of records. It fails when agencies cannot maintain custody of the entities they care about as data flows across internal systems, commercial feeds, and classified holdings. That is where entity resolution provides the connective tissue, maintaining a consistent identity for the same company, person, vessel, or facility across foundational and dynamic sources.

From Data to Seizure: The Oligarch Yacht Case

Kesecker shares a compelling example. When authorities seized a massive yacht in San Francisco, investigators needed help connecting ownership through shell companies. The investigative team traced ownership to a law firm, then used social media intelligence to find a photo of the oligarch on the boat. Combined with ownership data resolved through SENZING® entity resolution, the team built a prosecution packet that supported both seizure and sanctions.

Koczan offers another scenario: in a counter-drug investigation, entity resolution connected a suspect’s cell phone to corporate ownership and US commercial real estate, helping to surface companies producing fentanyl precursor chemicals and the properties they control near US airfields. “You can tell where they’ve been, which companies they’ve been corresponding with, which US commercial real estate pieces of property they’ve been at,” Koczan notes, describing how knowledge graphs lay these connections alongside classified holdings in a single view.

From Point-in-Time to Continuous Vetting

The panel emphasizes shifting from periodic checks to perpetual monitoring. Traditional vetting approaches take snapshots, but threats evolve continuously between checks. “With de minimis shipments, we would notice: why is there this one person that’s the director of 40 different companies?” Koczan explains. Continuous entity resolution identified the same individual serving on the boards of dozens of small exporters, each shipping just below de minimis reporting thresholds. Looked at one by one, they were unremarkable. Seen together, they formed a coordinated network rather than 40 unrelated firms.

Koczan shares another example: a UK-based company that banked with a prominent British bank had an Iranian national on its board. “You would have never realized that there was a connection to Iran, but combining the Kpler data with that ownership data” made the relationship visible.

The Reality: All Data Is Messy

The panel concludes with frank Q&A acknowledging data quality challenges. One attendee observes, “There’s no such thing as good data. All data is nasty, evil, rotten, and trying to break your system.” Kesecker adds that serious cleaning and entity resolution can create an optics problem. When Kpler and similar providers deduplicate their files, a headline number might drop from 2 billion to 1.2 billion records.

On a slide that looks weaker. Operationally, though, that “smaller” dataset is better. It reflects fewer duplicates and more accurately resolved entities, which is exactly what targeting and analytics teams need.

Video Highlights

• 03:38 – Data interoperability and entity resolution for national security
• 12:22 – Russia-Ukraine behavioral patterns: circular ownership anomalies
• 21:57 – Seizing an oligarch’s yacht using connected data
• 24:08 – Continuous vetting: 5-6M monthly ownership changes
• 27:30 – Iranian national on UK company board example
• 34:50 – Reality check: dealing with messy data quality

Ready to Connect Your Data for Mission-Critical Decisions?

See how Senzing entity resolution can help your organization transform isolated datasets into actionable intelligence.

Watch the full panel above, or Schedule a Demo.