I have an enormous collection of ROMs for various emulators. There are people who catalog the available ROM images for a given platform and make DAT files that list them. Each ROM has a checksum in the DAT file so if you have a file and you don't know what system it goes with or it has be renamed to something wrong, the DAT file will tell you what you've got.
Unfortunately, the original purpose for these DATs was to identify ROMs, so even versions of ROMs that were broken were listed in the DAT files. I personally use the DAT files to meter my collection, and I have no use for ROMs that will never work in any emulator. So when I use ClrMamePro to examine my collection for a particular system, it'll say that I'm missing x ROMs and I have to wade through the list of what's missing to see if there is anything there that I actually want. For some systems, I just download complete bundles of ROMs including all the hacks, overdumps, and broken versions. But for some systems, this is not an intelligent strategy. The N64 ROMs, for instance, comprise over 25 gigabytes - and at least 10GB of that is useless ROMs.
So I'm writing a parser that will filter the bad ROMs out of the DAT files and make my own versions. Then the ClrMamePro output will actually tell me what I want to know, and I won't have a mountain of gimpy ROMs in the file list when I play games on an emulator.
The question is: What to filter? I'm definitely ditching the bad dumps, overdumps, and bad checksum ROMs. Should I keep the translations? Fixes? Alternate dumps? Cracks? Any advice? Brent? :)
Once I get the tool finished, I'll produce a current set of DATs based on the latest releases of GoodDATs and put them online.
I recommend you wipe your entire collection and replace it entirely with remaned duplicates of Fire Truck and Pachinko Sexy Reaction.
I took out all of the Bad Dump, Overdump, Trained, Fixed, and Old Translation ROMs out of my collection. Bad/overdumps are a gimme, but I took out the Trained and Fixed ROMs because they're modified for use with emulators, so they're not the actual ROM file from the game (and most aren't needed anymore since emu's keep getting better).
I kept tranlations because there are games I'd like to play that never came over here (Fire Emblem on NES for exmaple), but they tend to come in versions, so I only kept the latest ones.
I kept Alternate simply because they are actual dumps of games, even if they are only one small change from the original, they were for sale at one point.
I kept everything else, including Demo/hacked ROMs (which are in thier own folders). I find those interesting just to see what people have done w/ games, or on their own. However they can be a pain since some systems have hacks w/ many many versions and it's annoying to have all of them, but it's also a pain in the ass to weed through them all. There are so many.
But yeah, getting rid of what I mentioned above I think is the place to start. It makes the most sense to me.