Iron Mountain InSight Intelligent Document Processing whitepaper

Successful organisations can leverage both physical and digital information to make informed decisions, better understand customers, innovate and grow, all while quickly responding to audit requests.

Too much knowledge and information remain trapped because organisations can’t access, comprehend or use the data contained within documents. But even in organisations digitising documents, challenges remain in managing unstructured, semi-structured and structured data types in a wide variety of siloed application environments.

The gulf between those organisations effectively extracting, managing, and leveraging their information and those that don’t will continue to widen, as artificial intelligence (AI) and machine learning (ML) are rapidly making their way into every aspect of our lives. It’s an AI world, and successful organisations will harness this technology.

With Iron Mountain Intelligent Document Processing (IDP), you can:

• Quickly turn documents into information you can use by digitising, extracting, classifying and verifying information with speed and accuracy so you can make more informed decisions, and enhance customer service.
• Increase productivity by creating your own customised, automated document processing workflows, and/or leverage a team of AI experts, so you can innovate and grow.
• Reduce time, effort, cost and errors with intelligently extracted and classified data so you can speed time to discovery or audit response.

Iron Mountain InSight Intelligent Document Processing

The evolution of document processing

OCR helped to usher in the first wave of digital document management in a world that was still primarily driven by physical document handling. The conversion of paper documents to text-enabled enterprise document management solutions, however, the process and workflow revolved around storage and retrieval of documents. While an important step forward, teams needed to structure, enrich and tag the information.

Document and information management has grown, fueled by the efficiency and productivity gains that electronic records could provide, but it was the addition of machine learning (ML) that truly unlocked the opportunity. By adding intelligence, through AI and ML, a new generation of document management has emerged: Intelligent Document Processing (IDP).

Iron Mountain InSight IDP includes high-speed scanning, data extraction, classification and enrichment. Human in the loop (HITL) is available to manage exceptions and to refine and retrain AI models leveraged in the system. The document processing workflow can be customised for your specific workloads, applications and environment via managed services. Or, you can create your own document processing workflow in the low-code development environment with an AI model library, tools for labeling and training, monitoring and exception processing.

The Intelligent Document Processing platform

The Iron Mountain InSight IDP platform is designed to integrate into an organisation’s content management and document workflows, providing intelligence that spans from ingestion of information through to data visualisation and processing automation. Key elements of the platform include accuracy, efficiency, predictability and visibility — areas where an IDP solution needs to be able to scale — to meet not only the needs for today, but also the needs of the future.

The entire orchestration of the solution is broken down into six phases, Import & recognition, Classification, Data extraction, Enrichment, Human-in-the-loop, and finally Integration. Each of these steps is critical to the overall accuracy and effectiveness of the solution.

Import and recognition

The first step in orchestration is importing the documents, often transitioning them from physical to digital. With high-speed scanning or computer vision, teams can ingest information into the system. Metadata, and data about the document is collected and stored.

Classification

Once in the system, the classification task can begin. In this step, the document’s structure and content are classified and information is validated. If there are separator sheets tied to the document ingested, these are removed to help streamline/optimise the process.

Data extraction

In this step, document types are aligned with applicable ML models that have been trained on that document type, enabling the system to extract information directly from the document, including fields like names, addresses, dates, currency amounts, and more. As the models have been trained to know where the content lies within the document, automating the extraction. Intelligent annotation is used to help extract unstructured data and information that can be used to help in the secure training and optimisation of the machine learning models.

Enrichment

During the enrichment phase, the system adds structure, context and meta-data to information to make it more usable. The resulting enriched content can then enable enhanced automation, e.g. enabling the system to validate data using APIs, even across multiple documents.

Human-in-the-loop

While the capability and speed of ML models can aid in the automation of the process, humans in the loop can handle the exceptions, and perform quality control.

Integration

Finally, after the data has been ingested, properly categorised, organised, and validated, it is then ready to be integrated back into the Iron Mountain InSight Content Management platform, or other document or data repository.