Rating system to detect bad songs (detect skipping)

[Maketsi]

I'm not sure if somebody has already suggested this idea, but because there's already some statistical information about listened songs, it wouldn't
be too hard to implement at least the following two features:

1) ability to detect songs which users don't like (songs skipped before the end). kPlaylist already saves this completion status in percents in the database. For example, there might be some kind of rating system, where completing a song would increase its points by 3, and skipping the file would decrease its points by 1. "Hot files" have many points, and "these suck" have -50 points... Randomizer could take these points in the account too.

2) Admin interface: ability to delete/move these bad songs away, or get a list of them, so an adminstrator can make something about it.

It's really hard to detect and cleanup these old songs which nobody want's to listen anymore (or worse, have never wanted). Skipping the song is easy (remotes for winamp), but if you must come to write every song name down, it sucks.

[aba]

1) ability to detect songs which users don't like (songs skipped before the end). kPlaylist already saves this completion status in percents in the database. For example, there might be some kind of rating system, where completing a song would increase its points by 3, and skipping the file would decrease its points by 1. "Hot files" have many points, and "these suck" have -50 points... Randomizer could take these points in the account too.

Well, the database should contain this information as of now. Every track
which is played is recorded with a percentage. (history table.) By doing
a simple SQL you should be able to retrieve less popular songs. If you
choose 'My favorites', 'Music match' or 'Alltime hits', this data is being used for popular songs.

Åge

[Maketsi]

Ok, I made a couple of sql queries which will reveal those unwanted songs I wanted.
Here's one of them, which will display the top20 list of songs which have been played at least 5 times, but no more than 30% complete on average. Output fields are: song id, listening count, avg. percent, directory, and the filename.

select h.s_id, count(*) count, sum(h.dpercent)/count(*) rate, s.dirname, s.free
from tbl_mhistory h, tbl_search s
where h.s_id = s.id
group by s_id
having count > 4 and rate < 30
order by rate
limit 20;