SAP SE
File validation supported by machine learning

Abstract:

Matching documents of a first kind (e.g., credit memos) to documents of a second kind (e.g., invoices) includes training a set of feature models and an all-features model. The set of feature models can be applied to unmatched documents of the first kind to produce predicted documents of the second kind. The predicted documents can be used to identify candidate documents as candidates to be matched to the unmatched documents.