Given user-inputted data, match up their fields to fields in my database

When the user uploads a spreadsheet with headers, I parse it into JSON. Then, I need to match their fields with the fields in my database. For example, if they uploaded a fname and lname field, I'd recognize those, and concatenate them into the full name field in my database.

My idea is to write patterns based on regex, column name, and column position. For example, a Last Name has a header name that probably starts with "Last" or at least "L". The values probably have a regex pattern of [a-zA-Z]{2,20} and the previous column's values (presumably first name) probably follows the same regex pattern. I'll keep my regex pattern simple & scan a sample of the records to rule out edge cases like "Jo Ann O'hoolihan".

Does this approach sound reasonable? Has anyone already created something similar? Of course, I'll have the user confirm that the algorithm guessed correctly, but I find it strange how little information I've found googling about the pattern. Any ideas, comments, or sources welcome!