Ratings Script Idea

Download and get help for different MediaMonkey for Windows 4 Addons.

Moderators: Peke, Gurus

Big_Berny
Posts: 1784
Joined: Mon Nov 28, 2005 11:55 am
Location: Switzerland
Contact:

Post by Big_Berny »

You could modify my script, so that you save the point-values into a customfield. Let's say it's 1000 by default (at the beginning). After that by rating (choosing) a song this point-value will increase or dercrease. Now my script will (if you modify it a little) convert (into normal curve) the point-values into ratings and save them into the rating field.
Image
Scripts in use: Genre Finder / Last.fm DJ / Magic Nodes / AutoRateAccurate / Last.FM Node
Skins in use: ZuneSkin SP / Eclipse SP
AutoRateAccurate 3.0.0 (New) - Rates all your songs in less than 5 seconds!
About me: icoaching - internet | marketing | design
meatlump
Posts: 18
Joined: Mon Aug 15, 2005 6:47 pm
Location: Pennsylvania, US

Post by meatlump »

I find this a very interesting idea. I've often wished I had such a system for comparing/rating photos or graphic images. It does take a long time, but helps get around the problem of rating consistency.

Without such a system I find it difficult to decide whether to give 3.5 stars or 4. I also may rate songs differently depending upon my mood. A system of comparative rating over time smooths out any inconsistencies.

There is an alternative to the original mechanism that was hinted at in one post. It is not implemented in MM and would require a script or a lot of manual work. Use large numbers (0-2000) and use an adjustment algorithm like chess ranking. If you compare two tracks rated close together the changes would be small. If you compare two tracks with widely differing ratings and like the lower rated over the top rated, the adjustments would be large.

An example: Pick a 1200 rating over an 1150 rating and they change to 1205 and 1145. Pick a 1200 over a 1700 and they become 1300 and 1600.

This type of rating comparison is useful when you can not objectively assign ratings to individual entities. It allows you to make simple choices of A vs B and develop ratings over time and many comparisons.

If this were to be scripted, it would need to track the rating score, how many times it has been rated (more would be more accurate, less would need done more) and optionally a customized conversion of rating scores to MM ratings to be used in playlists, etc. (1800-2000 = 5 stars, 1600-1800 = 4.5 stars, etc.)

Another subject that has been touched upon is how to apply the rating range to your songs. Some people rate them based upon all music that they know, thus giving most tracks that they've chosen for their library scores of 4 or more. Others spread out their favorite music over most of the range.

We can debate the best way to make use of the range, but the best will be different for different folks based upon how they use MM. A single user with a small library can easily use the entire range, while a user who shares the library with others with different tastes will want to use 1 and 2 stars for music they may not want to hear much at all and most of their preferred music concentrated around 4.5 stars.

Thoughts? Discussion?
meatlump

There are three kinds of people in the world. Those who can count and those who can't.
Post Reply