Apparatus and method for automatically assigning general ledger codes to an accounts payable invoice

Abstract

A non-transitory computer readable storage medium has instructions executed by a processor to access an accounts payable invoice data set. Tuples of invoice line-items are gathered from the accounts payable invoice data set. A general ledger label is attached to each tuple to form labeled general ledger invoice line-item tuples. A single digit percentage of input information in the general ledger invoice line-item tuples is masked to form first revised general ledger invoice line-item tuples. A single digit percentage of input information in the first revised general ledger invoice line-item tuples is masked with mismatched values to form second revised general ledger invoice line-item tuples. A machine learning model is trained with the second revised general ledger invoice line-item tuples to form a trained general ledger machine learning model. An accounts payable invoice is received. The trained general ledger machine learning model is used to assign general ledger labels to items in the accounts payable invoice.