We introduce a novel type of text representation that preserves the 2D layout of a document. This is achieved by encoding each document page as a two-dimensional grid of characters. Based on this representation, we present a generic document understanding pipeline for structured documents. This pipeline makes use of a fully convolutional encoder-decoder network that predicts a segmentation mask and bounding boxes. We demonstrate its capabilities on an information extraction task from invoices and show that it significantly outperforms approaches based on sequential text or document images.

We present INDREX-MM, a main memory database system for interactively executing two interwoven tasks, declarative relation extraction from text and their exploitation with SQL. INDREX-MM simplifies these tasks for the user with powerful SQL extensions for gathering statistical semantics, for executing open information extraction and for integrating relation candidates with domain specific data. We demonstrate these functions on 800k documents from Reuters RCV1 with more than a billion linguistic annotations and report execution times in the order of seconds.