Compare all names and give suspected last names higher points if matching.
Especially helpful if one encounters names only added to in marriage with former name,
Helen Müller is later known as Helen Mueller-Luedenscheid if second name comes after marriage from a source using English keyboards and you do not want to normalize "ü" before comparison (still recommended)
;-))

For time-consuming but extremly helpful glossing over small errors in spelling, do the comparison via Levenshtein.
Hint: if performance of the optimized version (still vfp) found on the net is still to slow, it can be transcribed into a C-fll for a 4-6 times speedup.
Helpful if large datasets are used ;-))




>Hi,
>
>I have a generic "search" class that I am improving. Before a user can do an add to a table, it searches about 12 different ways and then presents the results to the user (to cut down on duplicates). I have always thought I should write an article on it as it has been quite effective at cutting down on dups.
>
>Revisiting it now and trying to improve one search. For background, this search works fine. I call it "first initial search".
>
>table contains: S. Stallone
>user enters: Sylvester Stallone
>
>class preps the search value by taking the first letter of the firstname and adds a period to do the search (lastname = "Stallone" and firstname = "S."). (of course, all searches case insensitive) Works fine.
>
>Where I am a bit stuck is trying to get an *optimized* search for the reverse of this:
>
>table contains: Scott Peck
>user enters: M. Scott Peck
>
>What I did (2001 vintage version): took just the first character of the firstname and searched for that (lastname = "Peck" and firstname = "M) - which actually does not catch the above.
>
>Reviewing the table though, I realize that most of the entries are like the above and so I need to change this.
>
>I could just strip the initial and do the search but I am wondering if there are other options - I have not used the LIKE operator in SQL. Is it optimized or can anyone suggest other fast ways to do this?
>
>Albert