How do you ensure data integrity when merging two large, disparate databases with inconsistent naming conventions?