Talk:Match Levels

From Dedupe

Individual/Contact/Person Level: I don't understand the argument that we would like to have record 2 (Rob Smith) and 4 (Bob Smith) matched and then remove one of them. These are most likely different persons, and "deduping" them would be a mistake!? /Bard

Interesting- I think you make a valid point, i'm not sure if I originally had a specific task/reason in mind. Maybe it would be more valid if further fields existed (e.g. date of birth or mothers maiden name). If I worked for a company last year and somehow found my way onto a promotional database- but now I work for a different company I would not want to be mailed twice (so the deduping should say this person is the same person and has probably just moved address or job?)

I agree that your scenario could be valid, and that 2 & 4 could certainly be misspelled duplicates. I also agree that to make this a valid example here, you would have to have more information to suggest it is a duplicate (but I suspect that you would need even more discriminating data than birth date for a name like Bob Smith ;-) /Bard

Even if they are the same person, keeping only one of the records might be a mistake. If only 1 of "Rob Smith" records has his phone number, and a different "Rob Smith" record is the only record with his address, then keeping only 1 of those records would be a mistake. If there are no conflicts between the 2 records (only not-available N/A marks), and they really are the same person, it would be best to merge all the data into one record. --68.0.120.35 16:02, 22 March 2007 (GMT)