US-20260127451-A1 - ROBOTIC PROCESS AUTOMATION VALIDATOR FOR DOCUMENT PROCESSING THAT INTEGRATES GENERATIVE ARTIFICIAL INTELLIGENCE MODELS

US20260127451A1US 20260127451 A1US20260127451 A1US 20260127451A1US-20260127451-A1

Abstract

A method is provided. The method is executed by an robotic process automation using a document processing engine. The method includes generate an extracted data prediction of a document by the robotic process automation using multiple distinct document processing operations of the document processing engine. The method includes automatically validating the extracted data prediction by the robotic process automation cross-checking the extracted data prediction with a validator panel of the document processing engine. The method includes determining a confidence for the multiple distinct document processing operations based on the validating of the extracted data prediction.

Inventors

Luke PALAMARA
Alexandru Cabuz
Ioana Gligan
Tudor SERBAN

Assignees

UiPath, Inc.

Dates

Publication Date: 20260507
Application Date: 20241107

Claims (20)

1 . A method executed by an robotic process automation using a document processing engine, the method comprises: generate an extracted data prediction of a document by the robotic process automation using multiple distinct document processing operations of the document processing engine; automatically validating the extracted data prediction by the robotic process automation by cross-checking the extracted data prediction with a validator panel of the document processing engine; and determining a confidence for the multiple distinct document processing operations based on the validating of the extracted data prediction.
2 . The method of claim 1 , wherein the extracted data prediction comprises a deterministic estimation of data extracted from the document.
3 . The method of claim 2 , wherein the data comprises an identified value, a target, a purpose, an output, an amount, or an answer.
4 . The method of claim 1 , wherein the document processing engine receives and processes a plurality of documents including the document to produce the extracted data prediction.
5 . The method of claim 1 , wherein the multiple distinct document processing operations comprises two or more of text and character scanning, table exporting, automatic document recognition processing, document format converting, optical character recognition processing, and image processing algorithms.
6 . The method of claim 1 , wherein the validator panel comprises one or more document processing validators integrated with language model (LLM), machine learning (ML), artificial intelligence (AI), document processing, or deterministic data extraction code.
7 . The method of claim 1 , wherein the document processing engine assigns a trusted status to each document processing validator of the validator panel that agree with the extracted data prediction to improve an accuracy of the document processing engine.
8 . The method of claim 1 , wherein the robotic process automation increases the confidence of the multiple distinct document processing operations when the extracted data prediction is valid.
9 . The method of claim 1 , wherein the robotic process automation identifies with a flag the extracted data prediction and the multiple distinct document processing operations when the automatically validation of the extracted data prediction determines a discrepancy between the validator panel and the extracted data prediction.
10 . The method of claim 9 , wherein the flag triggers a subsequent review and feedback process to enable the validator panel to learn and to improve an accuracy of the document processing engine.
11 . A computer program product of a robotic process automation using a document processing engine, the computer program product stored on a non-transitory computer readable medium and executable by one or more processors to cause operations comprising: generate an extracted data prediction of a document by the robotic process automation using multiple distinct document processing operations of the document processing engine; automatically validating the extracted data prediction by the robotic process automation by cross-checking the extracted data prediction with a validator panel of the document processing engine; and determining a confidence for the multiple distinct document processing operations based on the validating of the extracted data prediction.
12 . The computer program product of claim 11 , wherein the extracted data prediction comprises a deterministic estimation of data extracted from the document.
13 . The computer program product of claim 12 , wherein the data comprises an identified value, a target, a purpose, an output, an amount, or an answer.
14 . The computer program product of claim 11 , wherein the document processing engine receives and processes a plurality of documents including the document to produce the extracted data prediction.
15 . The computer program product of claim 11 , wherein the multiple distinct document processing operations comprises two or more of text and character scanning, table exporting, automatic document recognition processing, document format converting, optical character recognition processing, and image processing algorithms.
16 . The computer program product of claim 11 , wherein the validator panel comprises one or more document processing validators integrated with language model (LLM), machine learning (ML), artificial intelligence (AI), document processing, or deterministic data extraction code.
17 . The computer program product of claim 11 , wherein the document processing engine assigns a trusted status to each document processing validator of the validator panel that agree with the extracted data prediction to improve an accuracy of the document processing engine.
18 . The computer program product of claim 11 , wherein the robotic process automation increases the confidence of the multiple distinct document processing operations when the extracted data prediction is valid.
19 . The computer program product of claim 11 , wherein the robotic process automation identifies with a flag the extracted data prediction and the multiple distinct document processing operations when the automatically validation of the extracted data prediction determines a discrepancy between the validator panel and the extracted data prediction.
20 . The computer program product of claim 19 , wherein the flag triggers a subsequent review and feedback process to enable the validator panel to learn and to improve an accuracy of the document processing engine.

Description

FIELD The present invention generally relates to robotic process automations (RPAs) for document processing, and more specifically, to document processing validators for document processing that integrates generative artificial intelligent (AI) models. BACKGROUND Companies have thousands of paper documents. To electronically store these paper documents, each paper document must be scanned. Scanning the paper documents creates a repository of thousands of scanned files that are unmarked and unorganized. Note that each paper document is scanned by a scanner or a printer into a flat image that has no markings therein (e.g., a scanned files). Note further that the scanned file from the scanner or the printer is assigned a randomized alphanumeric name that offers no indication of a flat image type or items, fields, and information within the of the flat image. In turn, companies have to dedicate significant resources and time to a scan, review, and validate the scanned files. Conventional scanning, reviewing, and validating includes scanning a paper document to create a scanned file in a database, individually opening and reviewing the scanned file to identify the information therein, and validating the information. Conventional validating can further include individual annotations of items, fields, and the information within that scanned file while the scanned file is being reviewed. Each step of the conventional scanning, reviewing, and validating (conventional document processing) is laborious and tedious, as well as extremely time consuming. For example, because conventional document processing requires human intervention for validating documents, conventional document processing can take five (5) minutes for an individual scanned document in a best case scenario. Further, the costs of and time-consumed by the conventional document processing are compounded when thousands of paper files are bulked scanned. Thus, a problem exists where companies are only able to perform conventional document processing of scanned files, as no automatic document processing solution is available that accurately validates the information of scanned documents. Accordingly, what is needed is an automatic document processing solution that does not require human intervention because the automatic document processing solution provides a sufficiently high accuracy and/or confidence on a quality of automatically extracted information. SUMMARY According to one or more embodiments, a method is provided. The method is executed by an robotic process automation using a document processing engine. The method includes generate an extracted data prediction of a document by the robotic process automation using multiple distinct document processing operations of the document processing engine. The method includes automatically validating the extracted data prediction by the robotic process automation cross-checking the extracted data prediction with a validator panel of the document processing engine. The method includes determining a confidence for the multiple distinct document processing operations based on the validating of the extracted data prediction. According to one or more embodiments or any of the method embodiments herein, the document processing engine can be implemented as an apparatus, a system, and a computer program product. BRIEF DESCRIPTION OF THE DRAWINGS In order that the advantages of certain embodiments of the invention will be readily understood, a more particular description of the invention briefly described above will be rendered by reference to specific embodiments that are illustrated in the appended drawings. While it should be understood that these drawings depict only typical embodiments of the invention and are not therefore to be considered to be limiting of its scope, the invention will be described and explained with additional specificity and detail through the use of the accompanying drawings, in which: FIG. 1 depicts an architectural diagram illustrating an automation system according to one or more embodiments. FIG. 2 depicts an architectural diagram illustrating a robotic process automation (RPA) system according to one or more embodiments. FIG. 3 depicts an architectural diagram illustrating a deployed RPA system according to one or more embodiments. FIG. 4 depicts an architectural diagram illustrating relationships between a designer, activities, and drivers according to one or more embodiments. FIG. 5 depicts an architectural diagram illustrating a computing system according to one or more embodiments. FIG. 6 illustrates an example of a neural network that has been trained to recognize graphical elements in an image according to one or more embodiments. FIG. 7 illustrates an example of a neuron according to one or more embodiments. FIG. 8 depicts a flowchart illustrating a process for training artificial intelligence and/or machine learning (AI/ML) model(s) according to one or more embodiments. FIG. 9