RE: The PerfectMatch…

Louis LaBonté — Mon, 18 Dec 2023 14:19:41 GMT

The main problem I have with PerfectMatch is that it's very poor at distinguishing between different types of alphanumeric strings, and yet it replaces it shamelessly (in false seamless appearance) by an irrelevant so-called correspondence.

I don't remember the last example I had, but it might as well have had a file number like T-2201-25 and mistakenly recognize it as a date (22-01-25) and “auto-localize” it as “25 janvier 2022”. The problem with such a false correspondence is that the default penalties for “Auto-localization” and “Text replacement” are 0, so an incorrect replacement like that would still be considered a PerfectMatch (100%), which should never be the case.

I guess the main workaround is changing the applicable “Penalties” score (see below), but I wish it were at least set to 1 as absolute “out-of-the-box” initial default.

That being said, for those who want to avoid the same kind of bad suprises I get through PerfectMatch, here's how to change the Penalties setting: