Computer-based systems configured for automated activity verification based on optical character recognition models and methods of use thereof

A method comprising:

• receiving, by at least one processor, a digital image of a receipt from a receipt scanning tool associated with the user;
• wherein the receipt corresponds to a user payment initiated by the user to a receiver;
• wherein the receipt indicates receipt information comprises:
• a receipt payee identifier associated with a receiver of the user payment,
• a receipt amount associated with the user payment to the receiver, and
• a receipt date associated with the user payment to the receiver;
• utilizing, by the at least one processor, an optical character recognition model to encode the receipt information into transaction data so as to create a digital representation of the receipt;
• extracting, by the at least one processor, a receipt payee identification feature that represents the receipt payee identifier based at least on the receipt data;
• extracting, by the at least one processor, a receipt amount identification feature that represents the receipt amount based at least on the receipt data;
• extracting, by the at least one processor, a receipt date feature that represents the receipt date based at least on the receipt data;
• generating, by the at least one processor, a receipt feature vector based at least on a combination of the receipt payee identification feature, the receipt amount feature and the receipt date feature;
• accessing in an electronic account of the user, by the at least one processor, a log of electronic transaction authorization request records comprising a plurality of electronic transaction authorization request records associated with a plurality of electronic transaction authorization requests to execute a plurality of user payments relative to the account of the user;
• wherein each electronic transaction authorization request record represents a transaction authorization request of a plurality of transaction authorization requests;
• wherein each transaction authorization request comprises:
• a request payee identifier data item associated with an entity associated with each transaction authorization request;
• a request amount data item associated with a transaction authorization request payment amount;
• and
• a request date data item associated with a date associated with each transaction authorization request;
• generating, by the at least one processor for each electronic transaction authorization request record, a transaction authorization request feature vector based at least on the request payee identifier data item, the request amount data item and the request date data item;
• inputting, by the at least one processor, at least the receipt feature vector and the transaction authorization request feature vector into a resolution machine learning model to generate a prediction that identifies a particular electronic transaction authorization request record as matching to the receipt captured in the digital image;
• wherein the resolution machine learning model is configured to:
• ingest the receipt feature vector,
• ingest the transaction authorization request feature vector, and
• measure a similarity between the receipt feature vector and the transaction authorization request feature vector by applying resolution classification parameters of the resolution machine learning model to the receipt feature vector and the transaction authorization request feature vector;
• wherein the resolution classification parameters of the resolution machine learning model have been trained based on error between:
• historical electronic transaction authorization request records, and
• historical digital images of historical receipts;
• wherein the particular electronic transaction authorization request record corresponds to the user payment initiated by the user to the receiver;
• determining, by the at least one processor, a difference between the receipt amount and the transaction authorization request payment amount based at least on the receipt data and the transaction authorization request of the particular electronic transaction authorization request record so as to determine a discrepancy between:
• an electronic transaction authorization request to execute the user payment, and
• the user payment to the receiver via the receipt;
• determining, by the at least one processor, that the transaction authorization request comprises a fraudulent payment quantity based at least in part on the difference;
• modifying, by the at least one processor, the user account to deny the transaction authorization request prior to the transaction posting to the user account based on the fraudulent payment quantity;
• and
• causing to display, by the at least one processor, an alert on a screen of at least one computing device associated with the at least one user indicative of the difference.