Sreejith Madhavan

Chief Operating Officer

Zerone Consulting

Cognitive OCR Handholds RPA In Managing Unstructured Data Inputs

Availability of right data at right time is integral for any business to streamline its operations and scale further. Today data (information) is ubiquitous and its role in enabling digital transformation in enterprises cannot be dismissed. However, the real challenge of data is its enormous presence in various forms and in vast quantities. To manually analyze and derive insights from both digital as well as other forms of data (such as paper) will take a toll on one’s business.

Technologies like Robotic Process Automation (RPA) increases the potential of digital data and allows businesses to fast-track their performance. This blog discusses the emerging RPA opportunities fueled by Optical Character Recognition (OCR) solution.

RPA and OCR: A brief overview

Robotic Process Automation (RPA) is a fast-growing technology that empowers businesses to deploy intelligent software systems or software “robots” that imitate human actions. Today RPA is applied in a large number of industries in which humans perform repetitive, mundane, high-volume, and highly transactional functions. RPA helps organizations in reducing the cost of labor-intensive tasks, thereby saving the capital and resources for higher value activities. The technology enables businesses to attain remarkable improvement in terms of quality, compliance, customer satisfaction and speed-to-market.

Optical Character Recognition (OCR) is a popular technology that enables conversion of typed, hand-written, scanned or printed text into an electronically usable format. OCR is the most common method of digitizing printed texts from invoices, paper records, business cards, passport documents and so on. The technology makes it easy for users to edit, search, store and display information online. OCR is key to multiple machine-based processes such as cognitive computing, text-to-speech conversion, machine translation and text data mining. Today, OCR is a field of research in pattern recognition and AI technologies.

The Problem

Present day businesses depend on digital information more than ever before. In spite of the widespread adoption of digitization, not all business operations are fully digital yet.

The BBC reports that the average UK office employee utilizes 45 sheets of paper per day. In many workplaces, paper is still a reality. There are several instances where enterprise operations depend heavily on analog inputs, including printed copies or scanned forms of documents, invoices, CVs, discharge summaries, or application forms. This happens due to logistical or environmental reasons as well as the inefficiency of organizations to become digitally interconnected, which in turn increases the operational expenditure of those enterprises.

Limitations of RPA

While technologies such as Robotic Process Automation helps in solving this challenge to a great extent, RPA software robots cannot completely work independently since they understand and interpret information in a rule-based manner. That means, to accomplish automation responsibilities, RPA robots need access to structured forms of data. Even the most basic of software robots can read out information from electronic documents and process the data using predefined rules and advanced algorithms.

Since a scanned document copy typically assumes the form of an image, it does not provide an easy-to-extract format. Just as humans, the RPA software robots confront the same challenge of copying and pasting texts from images. To overcome this obstacle, RPA’s software robots can be integrated with OCR technologies specialized in document recognition, data capture and language processing. Similar to how a human would type out textual information from a scanned image, robots will be able to identify data in scanned documents, with more speed and accuracy.

The Solution: Integrating OCR with RPA

By integrating advanced OCR technologies, RPA software robots can interpret scanned paperwork and convert data into digital formats with unprecedented efficiency. UiPath, the popular enterprise RPA platform states that the sophistication of interpreting scanned documents with the help of OCR varies depending on how the input data is structured.

For instance, there can be simple, structured documents that have less content, better image quality, and follow same writing style and format throughout. Here the level of complexity is less and UiPath supports affordable solutions from Google or Microsoft, required to address the situation. In cases where the documents contain more information, have lower image quality, use artistic fonts, or follow irregular formats, information extraction becomes a problem. Handling such scenario calls for trusted integrations. In the case of UiPath, the RPA platform partners with ABBYY FineReader (world’s leading OCR software). With the help of ABBYY, UiPath deals with complex and unstructured scanned inputs. A case in point is the invoices from multiple vendors following different formats, submitted to the same buyer.

It is not necessary that scanned documents in the same business environment should always share the same structure. For example, multiple invoices may contain same type of information, but in a different range of positions. To incorporate this kind of data inputs into their automated processes, UiPath uses ABBY’s FlexiCapture, where a scanned document is put through a series of technical steps to make it ready for the RPA software robots to work on.

Advantages of OCR-RPA integration

Advanced character recognition tools help in classifying complex, semi-structured documents based on their scope. The documents pass through a series of steps before information extraction, which helps in creating templates to individually categorize the extracted data. Abby’s FlexiCapture allows to apply a monitoring tool to get relevant analytics and ensure that the system is up and running. It is possible to involve manual verification of extraction errors at this stage, to avoid any future faults. 

In short, OCR makes way for a seamless interaction with RPA in transforming unstructured data into structured data. Regardless of the opportunities that digitization provides, paper is likely to remain an essential part of business practices in the foreseeable future. At the same time, challenges of working with printed or scanned documents incur heavy operational and resource expenditure for organizations. Applying an intelligent combination of key technologies will empower businesses to unlock the potential of automation. The integration of OCR and RPA platforms streamlines the day-to-day operations of companies that rely on printed documents.