| CPC G06F 16/51 (2019.01) [G06F 16/258 (2019.01); G06F 40/103 (2020.01); G06F 40/123 (2020.01); G06F 40/174 (2020.01); G06V 30/19 (2022.01); G06V 2201/10 (2022.01)] | 8 Claims |
|
1. A method of enhancing compatibility of documents, extracting data to populate a database on a computer server, and training statistical models with data on the database, the method comprising:
receiving, at a computer hardware processor, two or more first documents in a first format, the first format comprising a format that the computer hardware processor cannot use to extract data to populate the database,
wherein, the two or more first documents are stored on the computer server;
converting, using the computer hardware processor to perform optical character recognition (OCR), the two or more first documents from the first format into a second format, the second format comprising a format that the computer hardware processor can use to extract data to populate the database;
populating, using the computer hardware processor, the database with data from the two or more first documents in the second format;
training, using a graphics processing unit (GPU) and the database, a first statistical model, the first statistical model comprising a deep learning artificial intelligence (AI) system, to determine a reliability of a submitter of a document;
training, using the GPU and the database, a second statistical model, the second statistical model comprising a deep learning AI system, to determine a prediction for a field in a training document received from an entity and compare the prediction for the field to a first entry found in the training document as it was received from the entity;
receiving, at the computer hardware processor, one or more second documents from a party;
wherein, when the one or more second documents are in the first format, converting, using the computer hardware processor to perform OCR, the one or more second documents from the first format into the second format;
running, using the computer hardware processor, the first deep learning AI system to determine a reliability of the party;
when the deep learning AI system determines the party is reliable, running, using the computer hardware processor, the second deep learning AI system to determine predictions for fields in the one or more second documents and to compare the predictions for the fields to second entries found in the one or more second documents as received from the party;
when determining that the second entries are accurate, extracting, using the computer hardware processor, data found in the one or more second documents, and populate the database;
when determining that the second entries comprise an error, notifying, using the computer hardware processor, the party with a suggested correction to the error:
receiving, at the computer hardware processor, feedback from the party comprising a confirmation of the suggested correction, another entry, or a reversion back to an original entry;
storing the one or more second documents on the computer server; and
extracting, using the computer hardware processor, data found in the one or more second documents, and populating the database;
when the deep learning AI system determines the party is not reliable, notifying the organization to facilitate preventative measures to avoid loss to the organization.
|