For series I think that a script could do a good job by matching season/episode number in the file title and from metaData on OS.
Let use
http://www.opensubtitles.org/en/subtitl ... -rhythm-en as an example:
we've got this file/hash (I wont go with the multiple name for a same hash in this but the script could use them) :
A.Gifted.Man.S01E09.480p.HDTV.x264-mSD.mkv .................................. ok
A.Gifted.Man.S01E09.HDTV.XviD-LOL.avi .......................................... ok
A.Gifted.Man.S01E09.720p.HDTV.X264-DIMENSION.mkv .......................... ok
A.Gifted.Man.S01E09.hdtv-lol.avi .................................................... ok
A.Gifted.Man.S01E09.HDTV.XviD-LOL.avi ........................................... ok
MONEYBALL (2011) BRRip XviD-FTW.avi ............................................ delete
Burn.Notice.S05E02.Bloodlines.HDTV.XviD-FEVER-GTtvSD.avi ................... delete
Real Steel (2011) DVDRip XviD-3LT0N.avi ........................................... delete
Sherlock Holmes A Game Of Shadows (2011) TS XviD-DTRG.avi ................. delete
Game Time (2011).avi ................................................................. delete
Hawaii.Five-0.2010.S02E05.HDTV.XviD-LOL.avi ................................... delete
hawaii.five-0.2010.206.hdtv-lol.avi ................................................. delete
Hawaii.Five-0.2010.S02E07.HDTV.XviD-LOL.avi ................................... delete
Hawaii.Five-0.2010.S02E09.HDTV.XviD-LOL.[VTV].avi ............................ delete
Odarennyj.chelovek.(1.sezon.09.seriya.iz.16).2011.XviD.WEB-D ................ well delete but maybe it shouldn't anyway it's the right season episode just in a unusual format for me...
so just this 'simple' script would remove all the series subtitle that are linked to a movie file hash, all the movie sub that are link to series file hash and all the wrong episode in a series, the only problem that stay is movie sub link to another movie file hash and series sub that are linked to same season/episode of another series file hash.
another movie to see how it could help :
http://www.opensubtitles.org/en/subtitl ... ctivity-en
except the first file I don't think there is event one other hash that fit the sub but as the download number is almost 60000 there is no chance a software could pick the right sub from xml-rpc.
I think this would safely take care of a high % of wrong hash, but I see no way of fixing safely the movie-movie case other than human reporting. this kind of reporting could be made easier by adding an 'X' to each file/hash and developing an helper tool for admin to check them easily, but that's asking a lot of developing time.