US 12,254,711 B2
Methods and systems for automatic document pattern recognition and analysis
Hashwanth Maripina, Hyderabad (IN); Mohit Bansal, Hyderabad (IN); Debanwesh Bose, Hyderabad (IN); Harshita Srivastava, Hyderabad (IN); and Aditya Ankur, Hyderabad (IN)
Assigned to HighRadius Corp., Houston, TX (US)
Filed by HighRadius Corp., Houston, TX (US)
Filed on Jul. 26, 2022, as Appl. No. 17/873,915.
Prior Publication US 2024/0037975 A1, Feb. 1, 2024
Int. Cl. G06V 30/418 (2022.01)
CPC G06V 30/418 (2022.01) 19 Claims
OG exemplary drawing
 
1. A method, comprising:
receiving, using a computing system, a first document containing first remittance information;
analyzing, using the computing system, the first document to extract the first remittance information from the first document, wherein analyzing the first document comprises:
obtaining, using the computing system, one or more historical documents containing historical remittance information; and
correlating, using the computing system and one or more machine learning algorithms, the first document to the one or more historical documents to determine whether at least one historical document pattern of one or more historical document patterns of the one or more historical documents correlates to the first document, wherein correlating the first document to the one or more historical documents comprises:
generating, using the computing system, one or more historical vectors associated with the one or more historical document patterns;
clustering, using the computing system, the one or more historical vectors into one or more document clusters based on a first similarity between the one or more historical vectors;
generating, using the computing system, a first vector associated with the first document;
correlating, using the computing system, the first vector with the one or more document clusters to determine a second similarity between the first vector and the one or more document clusters; and
based on a second similarity between the first vector and the one or more document clusters, determining, using the computing system, that the first document correlates to the at least one first historical document pattern contained within at least one of the one or more document clusters;
determining, using the computing system, that at least one first historical document pattern of the one or more historical document patterns correlates to the first document; and
based on a determination that the at least one first historical document pattern of the one or more historical document patterns correlates to the first document, extracting, using the computing system, the first remittance information from the first document based on the least one first historical document pattern.