Choosing the Right Capture Methodology
Introduction to Intelligent Capture
Content capture in the past decade has become an ubiquitous facet of ECM deployments in the enterprise. Whether you rely on integrated suites with embedded capture tools or best-of-breed capture systems, you’re using capture to some degree or magnitude. Capture ensurers all inbound content be uniformly digitized, inscribed with metadata or otherwise tagged and indexed, and ultimately, routed to appropriate departments and personnel in an organization.
Capture serves as the front desk to all ECM systems, through which all information passes. Today, the question is no longer if you capture, it’s what you capture, how you capture it, when you capture it, and who performs the capturing. What we are talking about here is the degree of capture maturity in your organization.
Today’s generation of capture functions as if it has a brain. This generation of capture, aptly known as intelligent capture, uses algorithms and advanced data extraction software to “read” and validate captured content through context and applied operational rules – as if it were thinking. It’s responsive, accurate, and capable of mass workloads.
In a world fraught with risk of fraud, intelligent capture learns, improving recognition rates and accuracy levels over time – increasing the value of its results and your investment.
In the Beginning
Before capture systems had any semblance of intelligence, we first gave them a way to see using Optical Character Recognition (OCR). This typically consisted of and was restricted to the use of document templates using designated areas for data extraction. This required placing data in specific data fields of a form or document, in order for it to be captured, extracted, and saved. Early form template-based OCR carried the weakness of inflexibility of data type, laced accuracy, and necessitated a significant amount of manual interaction, which made it slow.
Intelligent capture quickly evolved to include memory. Now, keywords, mapped fields, or templates are “remembered” by the capture system and recalled automatically when needed; the capture technology was learning. But this process was too contrived to be real learning, and favored brute force memorization direct to a database, as opposed to fluent, adaptive thought and recall.