3 ms·
Yeah, but then you get into disambiguation hell. My first serious coding project was an insanely ambitious attempt to do something like this with CH data. Great
by specproc 2y ago
Yeah, but then you get into disambiguation hell. My first serious coding project was an insanely ambitious attempt to do something like this with CH data. Great punt for a Masters thesis, but not the sort of thing I'd base proper research on.
Love the look of the library they used to do this, in considerably better shape than my attempt.
https://github.com/ribenamaplesyrup/sugartrail https://github.com/ribenamaplesyrup/sugartrail
- cjk2 2y agoNot research, but I spent 5 years writing matching algorithms for data for financial companies which was entered from random crappy mainframes driven by the lowest bidding data entry workers. This tied together the financial relationships of millions of people. The dumber it was the measurably more accurate it was. Hell even some of it used soundex scoring and a naive z-test to work out the confidence that two names were matching.