Fuzzy matching

Fuzzy matching lets DinMo compare values that are similar but not exactly identical. It is useful for secondary attributes that can contain typos, formatting differences, or small variations.

In DinMo, fuzzy matching is part of matching rules and should be used conservatively.

When to use fuzzy matching

Use fuzzy matching for secondary attributes such as:

  • first name

  • last name

  • company name

  • address-like attributes

Do not use fuzzy matching as the only reason to merge records. A fuzzy criterion must be anchored by at least one exact criterion in the same rule.

Anchored fuzzy matching

Anchored fuzzy matching means fuzzy comparison is allowed only inside a rule that also contains exact evidence.

For example:

Rule
Why it is safer

Exact email + fuzzy last name

Email creates a strong match block, then last name tolerates small variations.

Exact phone + exact country + fuzzy first name

Phone and country anchor the match before name similarity is considered.

This is safer than comparing every name against every other name.

Match types

DinMo supports these match types:

Match type
Use it for

Exact

Stable identifiers that should match after standardization.

Fuzzy Medium

Moderate similarity on secondary values.

Fuzzy Strong

More conservative similarity on secondary values.

Review fuzzy rules

After adding fuzzy criteria, review:

  • rule applicability

  • valid match rate

  • conflict rate

  • sample matched profiles

  • unusually large clusters

Use Review and monitor before relying on fuzzy matches downstream.

Last updated