Address matching isn't quite exactly related to the topic, but of course it's related to name matching somehow. Because of the scale of the address matching topic I will not discuss it on this BLOG and rather have a little bit discussion on name match.
Regarding matching on initial, it's quite useful when matching on household level; when you have father/son relationship, when they share the same first name.
For each name part we can have the following cases if we using fuzzy matching:
1. Exact match
2. Nick name match
3. Fuzzy match
4. Match on initial
5 No match
all cases are listed according their matching weight, actually you can have it differently;for surname we don't do match on first letter(case 4) and nickname(case2); also there could be cases when surname could change, in case of marriage.
Each name part itself has matching weight; for example match on surname has higher weight than match on first or second name, not sure about to compare weight of first name and second name against surname.
That mean we having here a scoring engine with possible tuneup parameters.
That was my approach.
Also a word about cleansing function used by Ira.
I think it's kind of so called matchkey generator and it take me back to the times when no fuzzy matching was used but we were using matchkey to do our match.
For example keeping first 3 letters and removing wovels and doubles.
Match key generation could be another topic of discussion, but I would stop here.
Thank you again for the interesting discussion and useful script.