Skip to main content

Entity Resolution — Animation Concepts

Four ways to visualize how Senzing entity resolution works. Each is self-contained, on-brand (Obsidian & Ember), and fully editable.

Concept 1 — Single match: record-to-record vs. Entity Centric Learning

One record. Two methods. One match missed.

Why entity resolution must compare against the whole entity — not record by record.

Record-to-Record Matching
CRM
NameSue JonesAddress—Phone—
BANK
Name—Address47 Oak StPhone—
TELCO
Name—Address—Phone312-555-0147
Comparing new record to each record…
NEW
NameS. JonesAddress47 Oak StPhone312-555-0147
× name only × address only × phone only
× No match — duplicate created
Entity Centric Learning
CRM
NameSue JonesAddress—Phone—
BANK
Name—Address47 Oak StPhone—
TELCO
Name—Address—Phone312-555-0147
Resolved entitySue Jones
NameSue JonesAddress47 Oak StPhone312-555-0147
Comparing new record to the whole entity…
NEW
NameS. JonesAddress47 Oak StPhone312-555-0147
✓ Match — 1 entity

More data → higher accuracy. Entity Centric Learning gets smarter over time.

Concept 2 — Records over time: duplicates pile up vs. one entity that learns

One identity, arriving over time.

Watch what record-to-record matching misses — and what Entity Centric Learning learns — as records keep coming.

NEW
Name—Address—Phone—
Incoming record 5 / 5
↓↓
Record-to-Record Matching
5 separate entities · 4 matches missed
CRM Sue Jones
BANK Susan Jones DUPLICATE
TELCO S. Jones DUPLICATE
INS Sue J. DUPLICATE
NEW Susan Jones DUPLICATE
Entity Centric Learning
1 entity · 5 records · 0 missed
Resolved entity Sue Jones
Known as
Sue Jones Susan Jones S. Jones Sue J.
Addresses
47 Oak St 12 Pine Ave 88 Elm Rd
Phones
312-555-0147 415-555-0199
✓ matched on “Susan Jones” — an alias learned back at t2

Record-to-record matching never learns — one identity shatters into duplicates. Entity Centric Learning gets smarter with every record.

Concept 3 — Resolution that sharpens with every record

Entity resolution that sharpens with every record.

The person never changes. Senzing’s resolution does — each new record gives a clearer picture of who they are.

EXT
Name—Address—Phone—DOB—
Incoming record 5 / 5
↓
Resolved entity Sue Jones 5 records · 1 entity
Known as
Sue Jones Susan Jones S. Jones ≈ Sue Jonse
Addresses
47 Oak St 12 Pine Ave
Phones
312-555-0147
Date of birth
1984-03-02
Resolution confidence 97%
What resolution just did Linked despite a typo — “Jonse” recognized as “Jones”.

The records describe one real person all along. Entity resolution is what gets smarter — turning scattered, messy data into one clear identity.

Concept 4 — Self-correction: one entity becomes two

When new evidence arrives, resolution corrects itself.

Two records looked like one person — until a third proved otherwise. Senzing revisits the earlier decision automatically.

Senzing re-evaluates and splits one entity into two — then links them as related. Self-corrected · 2 entities · 1 relationship
GOVGeorge Foreman Jr123 Main St · DOB 2001
↓
Resolved entity George Foreman Sr 123 Main St · DOB 1949
DMVGeorge Foreman123 Main St · DOB 1949
BANKGeorge Foreman123 Main St
Related · same address
Resolved entity George Foreman Jr 123 Main St · DOB 2001
BANKGeorge Foreman123 Main St · DOB 2001
GOVGeorge Foreman Jr123 Main St · DOB 2001

Most systems lock in their first guess. Entity resolution keeps every decision provisional — new data can revise what was already resolved, in real time.