Brain: neural engine for data capture

The ODR BRAIN engine is the heart of the automatic recognition system. The Brain project, springboarding from a long experience with the heuristic engine, gave birth to the start of a new system at the beginning of 2010 based on the most advanced stepwise refinement and spreadsheet optimization, such as the principle of maximum entropy, Fuzzy Artmap and Neural Network. Brain is a latest generation engine for the automatic extraction of information from images in documents. These frontier technologies guarantee absolutely unique results with recognition rates for the whole document above 50%, and with autovalidation percentages near 90% for the individual fields. Brain classifies the document based on a series of "rating" parameters that define its quality and reliability, and according to this evaluation it sends the autovalidated information (without human intervention) to the automatic entry module on the corporate accounting system or to quality control, should the document present uncertain elements or inconsistencies. Brain reacts to the corrections in real time, readapting the recognition matrices, while the custom autovalidation rules and the parameters of common use for its functioning are modifiable on the spot in any given moment in order to guarantee the continuity of the document's processing. A set of configurators and a cockpit for writing the rules using natural language make this new product complete.

Main Recognition and Classification Functions:

• Automatic recognition of the supplier
• Automatic definition of the rules associated with the supplier, with self-learning means
• OCR recognition and voting on the result
• Calculation of the document rating
• Routing of documents that present errors or inconsistencies to the validation module