Multi Invoice Document Process Using AI - Updated Version

The document discusses Multi-Invoice Document Processing using AI/ML to handle complex PDFs containing various document types. It outlines challenges in processing multi-page invoices and proposes a solution involving digitization, intelligent classification, and splitting PDFs for individual processing. The implementation resulted in an 80% reduction in processing time, enhancing productivity and customer satisfaction.

Uploaded by

sakhich1992

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as DOCX, PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

5 views4 pages

Multi Invoice Document Process Using AI - Updated Version

Uploaded by

sakhich1992

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as DOCX, PDF, TXT or read online on Scribd

You are on page 1/ 4

Multi Invoice Document Processing Using

AI/ML
Document types like invoices, receipts, ACORD forms, and others are often combined
into a single PDF file, with some documents spanning multiple pages. The process of
handling and extracting data from such PDFs is known as Multi-Invoice Document
Processing.

Key Challenges in Document Processing:

In today's fast-paced IT environment, ensuring high accuracy and efficiency in

document processing is crucial. To meet this challenge, we have implemented
advanced document understanding capabilities. However, a question arises: How do we
handle documents with multiple pages, each containing different types of invoices within
a single PDF? What is the most effective way to process such complex documents?

Solution to Mitigate the Challenge:

To address this issue, our approach will be as follows:

 Digitization of Raw PDF: First, we will digitize a single raw PDF containing
multiple document types. Some of these document types may span multiple
pages.

The image below illustrates a single PDF that contains multiple invoices:
 Intelligent/Keyword-Based Classification: The document will first be classified
using an intelligent or keyword-based classifier. If the confidence level for any
document type falls below a certain threshold, the document will be routed to the
action centre for further classification.
 Classifier Training Approach: To improve accuracy, the classifier will be
trained with individual PDFs that contain only one document type—regardless of
whether the document spans a single page or multiple pages. This focused
approach will enhance the classifier’s ability to accurately process and categorize
documents.
The image below illustrates a single PDF containing multiple document
types, which has been routed to the action center for further classification:
 Splitting the Raw PDF: Based on the classification results array and the
document bounds property, the raw PDF will be split into multiple PDFs using
activities from the PDF package.
The Extract PDF Page Range activity will be leveraged to separate the raw
PDF, which contains multiple document types, into individual, categorized PDFs.

 You can set the range as shown in the snippet below. The current item type is
UiPath.DocumentProcessing.Contracts.Results.ClassificationResult, which is
provided for reference and may vary depending on your scenario:
 Next, we can process the individual split PDFs using the Document
Understanding template provided by UiPath. This involves incorporating the
Machine Learning Extractor, which can be retrained and fine-tuned later to
improve model performance and accuracy using a classic approach.
 Depending on the business scenario, you may also choose to adopt the modern
approach for enhanced flexibility and efficiency.

Conclusion: - As a result, the business achieved an 80% reduction in processing time,

streamlining operations and eliminating time-consuming manual efforts. This allowed
employees to allocate more time to critical, value-driven tasks, improving productivity
and decision-making. Furthermore, the increased efficiency led to enhanced customer
and stakeholder satisfaction, as processes were completed faster and with greater
accuracy.