Smart retail analytics and commercial messaging

Abstract

A real-time fraud prevention system enables merchants and commercial organizations on-line to assess and protect themselves from high-risk users. A centralized database is configured to build and store dossiers of user devices and behaviors collected from subscriber websites in real-time. Real, low-risk users have webpage click navigation behaviors that are assumed to be very different than those of fraudsters. Individual user devices are distinguished from others by hundreds of points of user-device configuration data each independently maintains. A client agent provokes user devices to volunteer configuration data when a user visits respective webpages at independent websites. A collection of comprehensive dossiers of user devices is organized by their identifying information, and used calculating a fraud score in real-time. Each corresponding website is thereby assisted in deciding whether to allow a proposed transaction to be concluded with the particular user and their device.

1 . A computer-implemented method for examining behaviors behind payment transactions to assess and protect from financial losses caused by high-risk users, comprising:
 receiving, at one or more processors, a plurality of transaction records respectively corresponding to a plurality of transacting entities;   generating, via the one or more processors, a plurality of smart agent profiles, each of the plurality of smart agent profiles—
 corresponding to a respective one of the plurality of transacting entities, 
 reflecting a plurality of individual and expanded attributes of corresponding transaction record(s), 
 being structured as one or more rolling list(s) of vectors, each rolling list of vectors corresponding to a real-time or long-term interval and being divided accordingly across different memory locations, 
   pre-computing, via the one or more processors, a velocity count for and statistics of the plurality of individual and expanded attributes of each of the plurality of smart agent profiles;   forecasting, via the one or more processors, purchase behaviors for at least one of the plurality of transacting entities at least in part based on corresponding ones of the smart agent profiles and associated velocity counts and statistics;   receiving, at the one or more processors, a new transaction record corresponding to a first transacting entity of the plurality of transacting entities, the first transacting entity being associated with a first subset of the plurality of smart agent profiles;   generating, via the one or more processors, a computer fraud risk score for the new transaction record at least in part by—
 stepwise advancing the computer fraud risk score, one step-up or one step-down per attribute, over the attributes reflected in the first subset of the smart agent profiles, 
 adjusting the computer fraud risk score to reflect lesser risk if it is determined that the first transaction record conforms to at least one of the forecasted purchase behaviors, 
 adjusting the computer fraud risk score based on the output of a model trained on transaction records of the plurality of transaction records corresponding to at least two (2) of the plurality of transactional entities, the trained model being constructed according to one or more of: a neural network, case based reasoning, a decision tree, a genetic algorithm, fuzzy logic, and rules and constraints, 
   automatically outputting the computer fraud risk score from one network server to another as a machine determination relating to fraud risk.