Xoriant Azure OCR User Manual

Data Digitization using Azure OCR for a Leading Diagnostics Firm

SITUATION

The client is a leading global diagnostics company that manufactures a wide array of innovative medical diagnostic assays. The company had about 50,000+ scanned documents across 200+ document families including Tables, Label-Value pairs, Handwritten and strike-through Handwritten information. The client wanted a solution to digitize and index this information to equip scientists with real-time access to critical data records for research.

BENEFITS

• Achieved 99.5% accuracy levels in digitized documents with our data quality improvement efforts

• Equipped scientists with the ability to quickly search historical documents and visualize data for making product

improvements

• Minimized information search time from hours to seconds and improved investigation efficiency with a centralized data storage and retrieval system

• Enabled better quality management in the manufacturing process by equipping scientists with better analytical abilities

• Minimized resources needed for data mining with digital transformation, allowing them to work on critical investigations

• Increased customer satisfaction with accelerated product development and improvement in the overall release quality

SOLUTION

Xoriant Azure-based Smart Capture solution was used to make reports searchable for clinical scientists with limited IT knowledge. We performed data classification, data extraction, and the quality checks to convert the scanned documents into accurate digital records. The information was extracted from Tables, Label-Value pairs, Handwritten and Strike through handwritten documents spread across 200+ document families. Overall, 50000+ documents were digitized and indexed. Our efforts enabled scientists with access to manual records in digital format for seamless research and decision-making.

Our key contributions included:

• Handling document classification and data extraction using Azure OCR and AI-based software

• Performing quality checks for data accuracy

• Developing UI for quick document access

PARTNER: Xoriant WEBSITE: www.xoriant.com

LOCATION: Sunnyvale, California

ORG SIZE: 4000+ employees

PARTNER PROFILE:

Xoriant is a product engineering, software development and technology

services company, serving technology

startups as well as mid-size to large corporations. With 30 years in business, 4000+ employees and $240M+ in revenue we bring proven experience,

mature processes and expertise in

leading-edge Big Data, Analytics and Machine Learning technologies.

OFAC Compliance for a Fortune 100 Bank

SITUATION

The client is a leading Fortune 100 bank with a client base of over a million people across countries. As an international bank, the client had to stay compliant with countrywide regulations and standards to maintain their credibility as a trusted bank. Since the information in all annexure documents was scanned manually, there was a high risk of non-adherence to different regulations like OFAC, KYC, etc. The client wanted a faster, trustworthy, and automated solution that can overcome the challenges in complying with OFAC regulations which require thorough scrutiny of all the supporting documents involved in a particular transaction for pointing out suspicious or bad entity names.

BENEFITS

• Achieved 95% accuracy of compliance checks with the implementation of the solution

• Reduced manual tasks and operational costs by 30%

• Improved the productivity in document scanning and processing with an automated workflow process

SOLUTION

Xoriant Smart Capture solution helped the customer to scan all the transaction documents accurately to reduce manual efforts and errors. The key contributions included:

• Designing a flexible system that accepted input files in any file format including plain/OCR/HTML file seamlessly integrating

with the client’s third-party OCR tools

• Developing classification, decision-making and machine learning algorithms for address detections, categorizing text/data

into different entities and re-using them for future checks

• Implementing a UI-based training framework for applying incremental feedback for named entities

• Predefining a text template to support context-based training using machine learning algorithms for increasing the

scanning accuracy

• Providing the ability to locate and classify elements into pre-defined categories that helped in quick analysis of the

transactions/documents and improve efficiency