Talk:Jaro-Winkler distance

From Wikipedia, the free encyclopedia

[edit] adjwt

This makes no mention of the adjacent-weighting array (adjwt in Winkler's strcomp95.c). Is this additional function simply a fudge, or a fundamental part of the algorithm that should be mentioned? 82.2.93.1 (talk) 23:35, 9 December 2007 (UTC)

I've taken the bull by the other foot, and added a note to the article. 82.2.93.1 (talk) 23:36, 9 December 2007 (UTC)

As now clarified in the article, the adjwt array and string length compensation are not a part of the published versions of this algorithm, therefore are not reflected in the examples, even though they are in the reference code. 66.220.23.57 (talk) 16:38, 11 December 2007 (UTC)

[edit] Number of transpositions

I dont understand why t values are zero in (DWAYNE, DUANE) and (DIXON, DICKSONX) examples. I think that part is not clear. The character N and E in DWAYNE are found in DUANE, but in different positions. But it is written that t is zero not 2. If someone can clarify this, I will apreciate it. Bitkidoku (talk) 20:39, 13 May 2008 (UTC)

The definition of transposition given in the article is very unclear. In comparing CRATE with TRACE we have that all the letters match, but they are in a different order. By swapping the C and the T, we can turn CRATE into TRACE. Such a swap of two elements is called a transposition. In DwAyNE versus DuANE the matching letters are already in the same order D-A-N-E, so no transpositions are needed.  --Lambiam 02:19, 19 May 2008 (UTC)