SQL Left Fuzzy Join with Levenshtein Distance - sql

Question

I have two data sets from two different systems being merged together within SQL, however, there is a slight difference within the naming conventions on the two systems. The change in convention is not consistent across the larger data sample but normally requires one modification to match.
System 1 data
System 2 data
AA0330
AA_330
AA0340
AA_340
AA0331
AA_331
AA0341
AA-341
I have been using the below Levenshtein distance SQL function too fuzzy match and get the below result, however, end up with duplicate joins. How could I modify my code to mitigate this?
SELECT [System1].[mvmt1],
[System2].[mvmt2]
FROM [System1]
left join [System2]
ON dbo.ufn_levenshtein([System1].[mvmt1], [System2].[mvmt2]) < 2;
http://www.artfulsoftware.com/infotree/qrytip.php?id=552&m=0
Current output:
System 1 data
System 2 data
AA0330
AA_330
AA0330
AA_340
AA0340
AA_340
AA0331
AA_331
AA0341
AA-341
How can I make sure I only get one outcome from the join?

Answer 1

not the best solution , but you can compare first 2 character and last 3 character , if all the codes are following the same pattern (2 character on right and 3 digit at the end ):
SELECT [System1].[mvmt1],
[System2].[mvmt2]
FROM [System1]
inner join [System2]
ON left(mvmt1,2) = left(mvmt2,2)
and right(mvmt1,3) = right(mvmt2,3)

Answer 2

what about something like this (sorry about the poor formatting):
WITH Initial_Fuzzy_Join as(
SELECT [System1].[mvmt1],
[System2].[mvmt2] ,
dbo.ufn_levenshtein([System1].[mvmt1], [System2].[mvmt2]) as StringDistanceMetric
FROM [System1]
left outer join [System2]
ON dbo.ufn_levenshtein([System1].[mvmt1], [System2].[mvmt2]) < 2
)
SELECT mvmt1, mvmt2, max(StringDistanceMetric)
FROM Initial_Fuzzy_Join
Group by mvmt1,mvmt2