A better default scoring script for Amarok 2

[Karthik Periagaram]

Hi all, 

Me again. Here are the three changes that need to be made to the codebase to 
improve the default scoring algorithm.

1) File: src/Amarok.h line 90
Change function definition of computeScore() to the following:

inline double computeScore( double oldScore, int playCount, double 
playedFraction, int rating )
{
	double newScore;
	if( rating == 0 )
		rating = 5;
	if( playCount == 0 )
		oldScore = 0;
	newScore = ( ( 5 * rating + 0.5 * oldScore ) + ( 100 - ( 5 * rating + 0.5 * 
oldScore ) ) * ( playCount / ( playCount + 1 ) ) ) * ( 0.9 + 0.1 * 
playedFraction );
	return newScore;
}

Then,

2) File: collection/sqlcolllection/SqlMeta.cpp line 755
Change function call
setScore( Amarok::computeScore( score(), playCount(), playedFraction, rating() 
) );

3) File: collection/nepomukcollection/NepomukTrack.cpp line 316
Change function call
setScore( Amarok::computeScore( score(), playCount(), playedFraction, rating() 
) );

With these changes, the score algorithm change is complete. The new score will 
agree with your rating more and will approach 100 as the song playcount tends 
to infinity. Thus, it is a better algorithm to calculate how much the song is 
liked by the user. 

To test the change, it helps to set all scores to zero. A bit of mysql-fu 
should do it (though I don't know how to play with those things). I just nuked 
the contents of ~/.kde4/share/apps/amarok, but that might not be an ideal 
solution to most people. :)

Again, if you have questions, do ask me and I'll explain why this is awesome 
and how when plotted, this function looks really cool. :) Thanks all, for 
helping me put this together.

(markey: I figured #include <cmath> might not be needed after all- I replaced 
exp( -x ) with 1/(x+1) and it works equally fine!)

Karthik