Analysis of Documents Born Digital.

While traditional document analysis has focused on printed media, an increasingly large portion of the documents today are generated in digital form from the start. Such “documents born digital” range from plain text documents such as emails to more sophisticated forms such as PDF documents and Web documents. On the one hand, the existence of the digital encoding of documents eliminates the need for scanning, image processing, and character recognition in most situations (a notable exception being the prevalent use of text embedded in images for Web documents, as elaborated upon in section “Analysis of Text in Web Images”). On the other hand, many higher-level processing tasks remain due to the fact that the design purpose of almost existing digital document encoding systems (i.e., HTML, PDF) is for display or printing for human consumption, not for machine-level information exchange and extraction. As such, significant amount of processing is still required for automatic information extraction, indexing, and content repurposing from such documents, and many challenges exist in this process. This chapter describes in detail the key technologies for processing documents born digital, with a focus on PDF and Web document processing.