Use AI Models for approximate matches

paolochip
paolochip
Altair Employee
edited November 2024 in Community Q&A
Hi,
many many times I'm struggling with joining/lookup values from different data sources that have sometimes slighlty differences hard to be coded or predicted. One of the classical example is two customer lists where the same customer can have capital vs not, commas, spaces and other differences in its name in a very random ways (e.g. CustomerA ltd vs customerA, ltd. or Customer B vs BCustomer, etc..) . I was wondering if there are existing AI models that can be used in RM that can solve this issue.

Thanks,

Paolo
Tagged:

Answers

  • MartinLiebig
    MartinLiebig
    Altair Employee
    Hey,
    last time I did this with a customer we ended up using Levenshein Distance. One may use word embeddings for it if you want to use some 'AI'.
    Cheers,
    Martin
  • paolochip
    paolochip
    Altair Employee
    Hi Martin, thanks for your quick response, we will try and see

    Paolo