Document-based fraud detection

A computer system for facilitating detection of document-related fraud, the system comprising:

• a rules database configured to store fraudulent document detection rules;
• one or more processors;
• and
• a non-transitory memory storing instructions that, when executed by the one or more processors, cause the one or more processors to
• generate or update the fraudulent document detection rules at least by training a machine learning program using at least (i) first image data corresponding to a plurality of first physical documents, and (ii) fraud determinations corresponding to the plurality of first physical documents, wherein the plurality of first physical documents and the fraudulent document detection rules correspond to a first type of document, and wherein generating or updating the fraudulent document detection rules includes using the machine learning program to identify at least one counterfeit factor probative of whether the first physical document is a counterfeit document and at least one forgery factor probative of whether information entered into a field of the first physical document was forged,
• receive second image data corresponding to an image of a second physical document, the second physical document corresponding to the first type of document,
• determine, by applying the fraudulent document detection rules stored in the rules database to the second image data, that the second physical document is fraudulent based at least in part on a first analysis of the at least one counterfeit factor and a second analysis of the at least one forgery factor, and
• cause an indication that the second physical document is fraudulent to be displayed via one or more computing device user interfaces.

A non-transitory, computer-readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to:

• generate or update fraudulent document detection rules at least by training a machine learning program using at least (i) first image data corresponding to a plurality of first physical documents, and (ii) fraud determinations corresponding to the plurality of first physical documents, wherein the plurality of first physical documents and the fraudulent document detection rules correspond to a first type of document, and wherein generating or updating the fraudulent document detection rules includes using the machine learning program to identify at least one counterfeit factor probative of whether the first physical document is a counterfeit document and at least one forgery factor probative of whether information entered into a field of the first physical document was forged;
• receive second image data corresponding to an image of a second physical document, the second physical document corresponding to the first type of document;
• determine, by applying the fraudulent document detection rules stored in the rules database to the second image data, that the second physical document is fraudulent based at least in part on a first analysis of the at least one counterfeit factor and a second analysis of the at least one forgery factor;
• and
• cause an indication that the second physical document is fraudulent to be displayed via one or more computing device user interfaces.