Skip to main content
ABBYY Vantage also uses machine learning technologies. Training helps improve the document type and field extraction precision quality, which reduces the time needed to check the documents manually. The system “learns” by using the data obtained from manual review error correction (unless training has been disabled by the system administrator). Training occurs when correcting the field region on the document image. If the extracted data on the data form does not correspond to the actual data from the document, verify that the field region (the area on a document image from which the data is extracted) has been correctly defined. If not, correct it by navigating to the relevant field in the data form and use the cursor to select the correct region on the document image. The data from this region will be recognized and displayed in the field. After the manual review results have been saved, the changes you have made will be used to train the system, which will improve the recognition quality for all subsequent similar fields. Training is also carried out when correcting the document type. If the document type was assigned incorrectly, assign the correct type to the document. After you have saved the manual review results, the system will be more precise when assigning this document type. For more information about how Online learning works, see Online Learning.