I am not sure what's your problem. when you implement snapshot, you should shedule a purge job using dbms_defer_sys.purge to clean up snapshot log.
it supposes to verfiy all snapshots been updated then do the purge. my environment has 2 snapshots so far it works pretty "clean".