A Comparison of Text-Based Methods for Detecting Duplication in Scanned Document Databases.