I'd like to match and convert each sentence of the document to a dummy text in a given database (i.e. word file), that gives the closest hit probability. The dummy database can be a user generated list of texts, so it is neither definite nor set.

For example:
Sentence to be converted: Hello there, how are you doing?

A list of dummy (but not) text strings:
1. Hello here, how are you doing? (this is the closest match of course)
2. Hello ere, how are you doing?
3. ello here, how are you doing?

The converted result: Hello here, how are you doing?

Can it be done?