Intelligent diversification tool
AssigneeVisa International Service Association
InventorNaiju Kavumpurath
Life of this patent
Estimating the term…
Abstract
A machine-learning tool evaluates an acquirer's current portfolio and then develops a model portfolio that mathematically redistributes the effect of the current portfolio by suggesting business categories that would better serve the acquirer from a risk/reward perspective. The machine-learning tool is trained with model portfolios and then generates a suggested portfolio that incorporates the acquirer's current partners and supplements them with additional business categories that would improve the risk/reward metric. The machine-learning tool may also select specific businesses from within the suggested business categories for the acquirer to use in achieving the suggested improvement.
Claims14
A computer-implemented method of generating a recommended merchant category portfolio for an acquirer from a current portfolio of merchants of the acquirer using a machine-learning/artificial intelligence (ML/AI) tool, the method comprising:
- generating a request for transaction data for transactions processed by the acquirer involving merchants in the current portfolio from a transaction database;
- receiving the requested transaction data from the transaction database;
- characterizing the transaction data according to merchant category classifications resulting in a characterized dataset;
- applying the characterized dataset as input to the ML/AI tool;
- applying market data from a market database as input to the ML/AI tool via an AI engine of the ML/AI tool, the market data being derived from merchant transactions outside of the acquirer's current portfolio;
- at the ML/AI tool,
- generating an output of a recommended merchant category portfolio based on the characterized dataset and the market data, the recommended merchant category portfolio including recommended merchant category classifications,
- assessing the recommended merchant category portfolio based on one or more acquirer-specified criteria including at least one of fraud rate and chargeback rate of the merchant category classifications in the recommended merchant category portfolio, and
- if the recommended merchant category portfolio does not meet the one or more criteria, changing the input of the market data to the ML/AI tool via the AI engine and repeating the generating of the output of the recommended merchant category portfolio until the one or more acquirer-specified criteria are met;
- and
- receiving from the ML/AI tool the recommended merchant category portfolio at a user interface of the acquirer via an application program interface (API), the recommended merchant category portfolio including a percentage contribution of each merchant category classification in the recommended merchant category portfolio in graphical or tabular form.
A computer-implemented method of generating a recommended merchant portfolio from a current portfolio of merchants of an acquirer using an intelligent diversification tool that includes a machine-learning/artificial intelligence (ML/AI) tool, the method comprising:
- training the ML/AI tool using a plurality of existing acquirer portfolios;
- generating, at the intelligent diversification tool, a request for acquirer transaction data from a transaction database, the acquirer transaction data being transaction data processed by the acquirer involving merchants in the current portfolio;
- receiving the acquirer transaction data from the transaction database;
- characterizing the acquirer transaction data according to merchant classification codes of merchants in the acquirer transaction data resulting in a characterized dataset;
- applying the characterized dataset as input to the ML/AI tool;
- applying market data from a market database as input to the ML/AI tool via an AI engine of the ML/AI tool, the market data being derived from merchant transactions outside of the acquirer's current portfolio;
- at the ML/AI tool,
- generating an output of a recommended merchant portfolio based on the characterized dataset and the market data, the recommended merchant portfolio including recommended merchant category classifications,
- assessing the recommended merchant category portfolio based on one or more acquirer-specified criteria including at least one of fraud rate and chargeback rate of the merchant category classifications in the recommended merchant portfolio, and
- if the recommended merchant portfolio does not meet the one or more criteria, changing the input of the market data to the ML/AI tool via the AI engine and repeating the generating of the output of the recommended merchant portfolio until the one or more of the acquirer-specified criteria are met;
- and
- receiving from the ML/AI tool the recommended merchant portfolio at the acquirer via an application program interface (API), the recommended merchant portfolio including a list of recommended merchants in tabular form.
Dependencies are inferred from the claim text.