Review Request: MusicBrainz Tagger "restyling"

[Sergey Ivanov]

2010/11/2 Lukas <1lukas1 at gmail.com>
>
> Hi,
>
> Its nice to see such improvement :)
>
> Just a few things to add - It looks like entire result gets compared to entire data from track. It's hardly the best way to it, since the longer correct part is, the more similar results looks like. From the screen shots its obviously not - while title and artist remains the same, album is entirely different - so its not that close match.
>
> That's why, I think it is better to match field to field and use average to sort results down.
>
> This means title gets compared to title, album to album etc. Doing so would draw user attention to the field that has difference.
> To indicate match level/ration discrete rather than continuous colors might be used.
> * None (or default background color) - If the is no difference
> * Green - Difference is quite minor (like misspells, or fixing capitals)
> * Yellow - There is some similarity e.g. one is a substring of another ("top dance" and "Top Dance vol 2.0")
> * Red - Almost nothing is in common (line in screenshots album filed)
>
> To prevent color noise, only {border-bottom: 3px COLOR solid; font-weight: bold } (no bold for "none" color) an be used instead of entire background.
> Such way even color blind will know where to drag attention (bold)
>
> Thanks,
>
> Lukas

Results are compared to track exactly as you wrote - field to field,
but all this fields has different weights:

Title: 22
Artist: 6
Album: 12
Track number: 6
Track duration: 8

Probably I should change 'em.

Finder thresholds results by Its similarity level ( 70% minimum ), so
there is can not be any *Red* tracks. In case of discrete colors,
we'll get mostly all yellow and a couple of green ( 1..3 ) ( all
matching stuff case insensitive btw ). And I'm not agree with "* None
(or default background color) - If the is no difference", If we use
colors to indicate match ratio any user understand, that green is
good, red is bad and yellow ( orange ) is some transitive measure not
good not bad, something in between; But what uncolored items mean,
some results which program forgot to remove? May be I am very wrong,
but I think in this direction.

Good point about bottom line, I'll try. :)

--
Sergey