MrSinatra;299544 Wrote: > i'm just repeating myself now but for me and many others its wrong 50% > of the time AT LEAST. (meaning, half of what it IDs it IDs > incorrectly)
Within your sample data, perhaps. It's right almost all of the time for me. Go figure. MrSinatra;299544 Wrote: > it should be REPLACED with an effective methodology that does not ask > users to alter their tags. from a design POV, thats the holy grail, > and i've suggested a way in which it could be done. I've re-read this thread just to make sure I didn't skip something and failed to find this more effective methodology. You mentioned scanning for any tag containing the string "various artists", but this would also require retagging and doesn't seem any more effective at disambiguating data than the compilation tag. MrSinatra;299544 Wrote: > you certainly don't benefit from the VA logic. in fact, you have to set > a tag to defeat it. I said I used the explicit compilation tag for about 6 albums out of 300. That means the VA guessing algorithm has a 98% success rate. It saved me the work of explicitly tagging nearly 300 albums as either compilations or non-compilations, so I feel I've benefitted quite a bit. MrSinatra;299544 Wrote: > your premise is that "guessing" is the only way it could be done. i > reject that outright, and have an alternative that doesn't require > guessing. Any algorithm that makes a decision based on inadequate information is a guessing algorithm. Any method that uses "extra" data to figure it out is just a variation on the current behavior. -- CatBus ------------------------------------------------------------------------ CatBus's Profile: http://forums.slimdevices.com/member.php?userid=7461 View this thread: http://forums.slimdevices.com/showthread.php?t=47297 _______________________________________________ discuss mailing list [email protected] http://lists.slimdevices.com/lists/listinfo/discuss
