Self learning machine learning transaction scores adjustment via normalization thereof accounting for underlying transaction score bases relating to an occurrence of fraud in a transaction

A computer-implemented method of measuring transaction scoring data to detect one or more fraudulent transactions from among transactions corresponding to received transaction data therefor, the method being implemented in a computer system defining a fraud scoring engine comprising one or more processors configured to execute instructions stored on a non-transient computer-readable medium, the method comprising executing one or more of said instructions for:

• communicating with at least one first rules base and at least one second rules base each comprising at least one rule;
• in response to the communicating, applying the received transaction data to the at least one first rules base and the at least one second rules base;
• obtaining a normalization of the transaction scoring data,
• wherein the transaction scoring data is, for a first iteration of the received transaction data corresponding to at least a first transaction of the transactions, selected based on (a) an application of the first rules base to the received transaction data, in response to the application of the first rules base, as against an application of the second rules base to the received transaction data, generating a highest correspondence between the transaction scoring data and feedback, for the received transaction data, which is indicative of whether one or more portions of the received transaction data were fraudulent or (b) the application of the second rules base, as against the application of the first rules base to the received transaction data, generating a highest correspondence between the transaction scoring data and the feedback for the received transaction data, and
• wherein, for a sequential second iteration of the received transaction data, corresponding to at least a second transaction of the transactions, and in response to the application of the second rules base generating a highest correspondence between the transaction scoring data and the feedback for the received transaction data of the first iteration therefor, the second rules base comprises an initial rules base against which the received transaction data for the second iteration thereof is automatically applied.