World Classification

Question

I am trying to create something that takes a companies foreign GL accounts and maps it to my companies set of GL accounts.

GL account example - Foreign account names 12345 Market Rent Schedule 12344 Loss to lease 23456 tax

Local account names - 45678 Market-rent 45677 loss/gain to lease 56789 property taxes

so from a human perspective it would be easy to map these accounts and these are simple examples for the sake of explaining the problem domain. The general idea would be clean all the noise (numbers, symbols, and anything else unnecessary) but then I am not privy to a solution mechanism that would automate the process of linking each account. I believe it would be somewhere along the lines of using NLP to accomplish this but would appreciate more guidance if anyone knows a good source for this kind of problem preferably done in python.