Matching 2 columns from 1 table...fuzzy match an option?

[rhyatt]

Hi, I posted a week or so ago about trying to match up customer search terms with product names.  Using CONTAINS I am matching a decent amount of these, but I'd like to catch even more. I've been looking into fuzzy match, but I'm not sure how to use it for what I want.

Here's an example of what I'm talking about:

id	search term	product name	match
1	advil	advil gel 100	yes
2	advl	advil gel 100	no
3	advil gel 10	advil gel 100	no

For the purpose of what I'm doing I would like all of those to be considered a match.  Contains only counts the first one as a match since the term 'advil' is part of the product name. 

When i read up on fuzzy match I can't find an example that compares 2 columns in the same table.  I'm guessing I need to maybe union the table to itself and give each row a unique identifier and then somehow try and match the 2 columns from there?  But I haven't been able to get anything like that to work either. 

Help!

[mborriero]

Hi @rhyatt, the Fuzzy match tool has just one input. You need to union the field you  want to compare, then the option "Merge".

The workflow attached compare each word from the 2 source and give you a percentage of match score.

[rhyatt]

This looks very promising, but I'm having some trouble on my end.  With a similar fake data set (only 5 rows), I can only get 2 to show up in the final output when I believe they should all count as matches?

Can you look at this and see where I might be doing something wrong? 

Thanks!

[mborriero · Accepted solution]

I do not know the reason... but you have to do an UPPERCASE on one of the field.

I had your same issue yesterday, then I solved it using an uppercase function, but again, not sure why!

[rhyatt]

Ha, that's so strange.  But yep that did it, works now.  Thanks man!