[IMPALA-8752] Add Jaro-winkler edit distance and similarity built-in function

• Type: New Feature
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: Impala 3.3.0
• Labels: built-in-function
• Epic Color: ghx-label-6
• Assignee: Norbert Luksa
• Reporter: Norbert Luksa
• Created: 11/Jul/19 20:17
• Updated: 14/Aug/19 11:20
• Resolved: 14/Aug/19 11:20

Description

References:

• [Apache commons - JaroWinklerDistance |https://commons.apache.org/proper/commons-text/apidocs/org/apache/commons/text/similarity/JaroWinklerDistance.html]
• [Apache commons - JaroWinklerSimilarity |https://commons.apache.org/proper/commons-text/apidocs/org/apache/commons/text/similarity/JaroWinklerSimilarity.html]
• [Oracle - JARO_WINKLER[_SIMILARITY]|https://oracle-base.com/articles/11g/utl_match-string-matching-in-oracle]

Notable difference:

• With similarity, the Oracle version returns a normalized result ranging from 0 to 100.
• In the Appache version, null values result in exceptions.
• Apache rounds the values to two digits 

The scaling factor of the algorithm can be added as an extra/default argument.

Sub-Tasks

• Impala Doc: Document Jaro-winkler edit distance and similarity built-in functions: Closed