CLEANING A VERY LARGE DATABASE

I have a client who has the entire telephone directory of a large city in Mexico in a single database. For brevity, I have outlined the challenge below: 1. data stored in non-relational database, can be delimited and loaded into Informix. 2. fields used are similar to first, last and middle name, etc. 3. data is badly corrupted, for example, middle initial in the first name field, misspellings, abbreviations, etc. 4. client wants to clean and load data into an RDMS, possibly Informix. Are there any suggestions/strategies/third party tools recommended for this task? Mike.