258 UTL_MATCH

Overview

"Edit Distance" also known as "Levenshtein Distance "(named after the Russian scientist Vladimir Levenshtein, who devised the algorithm in 1965), is a measure of Similarity between two strings, s1 and s2. The distance is the number of insertions, deletions or substitutions required to transform s1 to s2.�

The Edit Distance between strings "shackleford" and "shackelford" = 2

The "Jaro-Winkler algorithm" is another way of calculating Edit distance between two strings. This method, developed at the U.S. Census, is a String Comparator measure that gives values of partial agreement between two strings. The string comparator accounts for length of strings and partially accounts for typical human errors made in alphanumeric strings.

Table 258-1 shows similarity values returned by Jaro-Winkler and Edit Distance

Security Model

The UTL_MATCH package runs with definer's rights. UTL_MATCH must be created under SYS. Operations provided by this package are performed with SYS privileges.

EDIT_DISTANCE Function

This function calculates the number of insertions, deletions or substitutions required to transform string-1 into string-2.

UTL_MATCH.EDIT_DISTANCE (
   s1  IN  VARCHAR2, 
   s2  IN  VARCHAR2) 
 RETURN PLS_INTEGER;�

SELECT UTL_MATCH.EDIT_DISTANCE('shackleford', 'shackelford') FROM DUAL;  
-------------
returns 2 

EDIT_DISTANCE_SIMILARITY Function

This function calculates the number of insertions, deletions or substations required to transform string-1 into string-2, and returns the Normalized value of the Edit Distance between two Strings. The value is typically between 0 (no match) and 100 (perfect match).�

UTL_MATCH.EDIT_DISTANCE_SIMILARITY (
   s1  IN  VARCHAR2, 
   s2  IN  VARCHAR2) 
 RETURN PLS_INTEGER;�

SELECT UTL_MATCH.EDIT_DISTANCE_SIMILARITY('shackleford', 'shackelford') FROM DUAL;
--------------
returns 82

JARO_WINKLER Function

This function calculates the measure of agreement between two strings.�

UTL_MATCH.JARO_WINKLER (
   s1  IN  VARCHAR2, 
   s2  IN  VARCHAR2) 
 RETURN BINARY_DOUBLE;�

SELECT UTL_MATCH.JARO_WINKLER('shackleford', 'shackelford') FROM DUAL;
--------------
returns 9.818E-001

JARO_WINKLER_SIMILARITY Function

This function calculates the measure of agreement between two strings, and returns a score between 0 (no match) and 100 (perfect match).�

UTL_MATCH.JARO_WINKLER (
   s1  IN  VARCHAR2, 
   s2  IN  VARCHAR2) 
 RETURN PLS_INTEGER;�

SELECT UTL_MATCH.JARO_WINKLER_SIMILARITY('shackleford', 'shackelford') FROM DUAL;
--------------
returns 98