Fitter-Fixer is a project mainly written in ..., it's free.
I started building a data cleanser, with emphasis on categorical data. The project is very incomplete.
This is a Visual Studio 2008 project. You can either open it in MSDN or run the executable in the bin directory. It creates a blank DB2 database, loads CSV's, and gives options to either detect taxonomic errors or proceed with duplicates. It then either proceeds to correct taxonomic errors using a classification tree for text matching and WordNet databases for word sense disambiguation, or duplicates to store extra rows in an additional table.
DISCLAIMER: The work is very rough; please ensure to make a backup before testing it on your dataset.