The Categorization Engine
FAQ: How does your categorisation work?
A very common question is: how does your categorization engine work?
The details are rather complicated but we can outline the overall process as follows:
Certain situations are algorithmically "hardwired". For instance, will some return debit notes be generated by the bank systems according to specific rules that are catched in this first step.
The categorization engine executes a rule system that uses all kinds of available information, i.e. the account holder name, the turnover amounts and purposes, the counterparty names, IBANs and types and even turnover-spanning information like "this counterparty appears as being the employer of the account holder".
If several different categories are possible/likely, we employ an approach to select the most likely alternative of the possible alternatives.
In the future, we will add additional components to get better estimations for categories that are extremely difficult by nature (e.g. money transfer from one private person to another with a purpose of "dance, the king is in the mood").
Last updated