แก้ไข

EDIT_DISTANCE_SIMILARITY (Transact-SQL)

Applies to: SQL Server 2025 (17.x) Azure SQL Database Azure SQL Managed Instance SQL analytics endpoint in Microsoft Fabric and Warehouse in Microsoft Fabric SQL database in Microsoft Fabric

Calculates a similarity score between two strings based on their edit distance. The score ranges from 0 (no similarity) to 100 (an exact match).

Note

  • EDIT_DISTANCE_SIMILARITY is generally available starting with SQL Server 2025 (17.x) CU11. On earlier SQL Server 2025 builds, the function is in preview. If it isn't enabled by default on your build, enable PREVIEW_FEATURES by setting it to ON.
  • EDIT_DISTANCE_SIMILARITY is available in Azure SQL Managed Instance with the SQL Server 2025 or Always-up-to-date update policy.

Syntax

EDIT_DISTANCE_SIMILARITY (
    character_expression
    , character_expression
)

Arguments

character_expression

An alphanumeric expression of character data. character_expression can be a constant, variable, or column. The character expression can't be of type varchar(max) or nvarchar(max).

Return value

int

Remarks

This function implements the Damerau-Levenshtein (Optimal String Alignment) algorithm. If any of the inputs is NULL then the function returns a NULL value. Otherwise, the function returns an integer value from 0 to 100. The similarity value is computed as (1 - (edit_distance / greatest(len(string1), len(string2)))) * 100.

Examples

A. Calculate edit distance similarity between two words

The following example compares two words and returns the EDIT_DISTANCE_SIMILARITY() value as a column, named Similarity.

SELECT 'Colour' AS WordUK,
       'Color' AS WordUS,
       EDIT_DISTANCE_SIMILARITY('Colour', 'Color') AS Similarity;

Here's the result set.

WordUK WordUS Similarity
------ ------ -----------
Colour Color  83

For more examples, see the Examples section in the Fuzzy string matching overview.