The apparently soon to be released Gridworks project from Freebase (http://blog.freebase.com/2010/03/26/preview-freebase-gridwor...) looks like it could go a long way to help with data cleansing, especially with grouping and batch editing similar mis-spellings and such.