A computer-implemented method for identifying pattern deviations in regulatory documents compliant with financial regulations

Patentradars sammanfattning

The invention provides an automated pipeline for identifying falsifications in financial regulatory documents such as annual and quarterly reports. Upon receipt, a document is digitized into a 2D grid representation, with content grouped by visual appearance into types (text, numbers, tables, footnotes) and then sorted into structured categories. The resulting 'specific regulatory document material' is compared via pattern-analysis algorithms against a machine-learning-trained corpus of predefined general regulatory document materials to extract deviations—such as font inconsistencies, misaligned columns, calculation errors, and irregular terminology—that collectively indicate a probability of falsification. Adjustable tolerance thresholds control alert sensitivity, and all processed materials are stored for retrospective automated auditing.

AI-genererad och redaktionellt processad.