in reply to De Duping Street Addresses Fuzzily
We are doing something simular in our company (for addresses is germany) as one small part of our jobs. Therefore the addresses get plausibilised (if I can say this in this way(?)) with some kind of technic simulare to the technics described in the posts above.
If you want accurate results, don't think it's fast done. Perhaps it could be cheaper to buy a complete solution.
Data that comes from humans contains all kind of errors and rubish you can and can not think of. Specially if it's a large mass of data (such as 50000000 addresses or else), there will be almost everything in your data - be aware.