Abstract: Systems and methods for automatically understanding, decomposing, extracting, validating and reformatting unstructured tabular information into intermediate structured representations of the information contained therein are described. No constraints are placed on the origin or format of these documents when originally submitted. Furthermore, no pre-created scripts are required to map the information contained in the submitted documents. The systems and methods of this invention generally comprise obtaining an electronic document, automatically analyzing and understanding the contents of the document, extracting information from the document, categorizing the information, and then creating an intermediate structured representation of the information contained therein. The intermediate structured representations may then be easily converted for use in a myriad of back-end systems.