Bill Talada - Monday, February 20, 2017 11:51 AM

As I've had to do this on many projects, I will share my past solutions.

I use a table of about 4000 FirstName-NickName pairs to find likely equivalents.  Is there a way I can share this file?

I take each row and look for possible matches to it since there could be several duplicates (many to many).

I give each match a reliability rating.  For example SSN+LastName is close to 100% while FirstName+DOB might be 80%.

I use Soundex matching but assign lower reliability ratings for those rows.

You cannot catch transposition and spelling errors easily except by looking at other column matches which result in lower reliability ratings.

My preference is to let users merge accounts based on human intelligence while processing my work queue.

There are no shortcuts in life.