Native Entity Resolution: what it is, how it works, and what it changes

What changed

Entity resolution has always been a solvable problem with the wrong architecture. Dedicated external ER engines are accurate but sit outside the investigation platform – requiring integration, a separate vendor, and a data movement problem. Rules-based matching breaks on data variability, scales badly, and requires re-engineering every time a new source is added.

DataWalk Native Entity Resolution changes the architecture. The engine runs natively inside DataWalk. No external system. No data movement outside the platform once data is ingested. For air-gapped and restricted environments, that’s not a preference – it’s the only viable architecture.

Resolved entities are immediately available for graph-based investigation. There is no separate integration step. That’s new.

How it works

The engine combines deterministic logic with phonetic encoding, fuzzy string similarity scoring, and geographic proximity matching. It handles typos, missing fields, and deliberately falsified data. Every match produces a human-readable audit trail. No black-box logic.

Smart blocking narrows candidate comparisons before matching begins, preventing exponential performance collapse at scale.

Delta-based processing means only new or modified records are evaluated on each run. Enterprise-scale updates run continuously without reprocessing the full dataset.

Network-aware resolution validates potential matches against real relationship context – shared accounts, devices, counterparties, addresses.

Analyst-in-the-loop control allows force-resolve or force-split when new intelligence surfaces. Every action captured: user, timestamp, justification. Fully reversible.

Performance

94.97% recall (deterministic matching)
260 million records, approximately 15 billion comparisons, in under one hour
New data sources fused dynamically – no model re-engineering required
Deployment in days, not months

Early Access Program

Native Entity Resolution is available to a select cohort of existing customers managing high data volumes. Participation includes co-implementation on your infrastructure using real data, direct access to the DataWalk engineering team, and roadmap input.

[LINK NEEDED: Early Access contact form] – Join Early Access → [this is an email for now]

READ MORE – WHAT’S NEW IN 5.2