understanding scoreDetails.authorMatchScore maximum value about gnverifier HOT 4 CLOSED

The score was made at first for internal use only to determine the best result. The way it works is by taking a very big 32bit number xxxxxxxx-xxxxxxxx-xxxxxxxx-xxxxxxxx and assign each score part a place of 1-3 bits on this number.
For example currently accepted names takes one bit 0 or 1, because it has only 2 states, and authorship has 8 states, so it takes 3 bits 000, 001, 010... 111

Every score than placed on the 32 bit binary number according to its priority, things on the left trump things on the right. The resulting 32bit number is used for sorting. So the final result of one matched result would look somethng like

00011000-1000000-01101000-00000000 -> 205547520

The algorithm was for intenal use, but then I found that people do not understand that there is author matching happening as well as other things, so I tried to make sense of this score in the output to show details of matching.

I need now to figure out how to assign weight to each scoring feature, so I can produce a more understandable (and reasonable) total number from 0.0 to 1.0 as a total score.

from gnverifier.

For botanical names it is true, the author score never goes to one. A name needs a matching year for both names to get to 1.
It is needed because for zoological names authors matching year perfectly should be higher than all other cases.

http://verifier.globalnames.org/api/v0/verifications/Bubo%20bubo%20Linn.%201758
http://verifier.globalnames.org/api/v0/verifications/Bubo%20bubo%20Linn.%201757

from gnverifier.

Ah!! Understood

So for my own needs we can say the botanical scoring ranges between 0 and 0.71428573? (so I can calibrate my results)

It's a strange maximum value. I suppose it's the result of some fraction, sqroot or whatever?

I ask just in case it might change in the future and break my code.

from gnverifier.

Score number is not added to the output (sortedScore field)

from gnverifier.