Automated quality assurance for document understanding systems

To process high-volume input data, such as the scanned images of publishersʹ book and journal collections, content understanding systems should run automatically, continuously, and without human attendance. Ensuring the output quality of such systems is a challenging task, however, and automated quality assurance techniques are thus essential to its success. The author discusses three automated QA techniques that were developed for Hewlett-Packardʹs Digital Content ReMastering system.