Systems and methods for adaptive identification of sources of fraud
A fraud detection engine is provided that analyzes transactions for fraudulent transactions. The transactions may include credit card or debit card transactions. The fraud detection engine may identify possible sources of fraud. The fraud detection engine may identify possible phony acceptors that masquerade as genuine merchants. The fraud detection engine may identify compromising points where accounts become compromised and are prone to fraudulent transactions thereafter. The fraud detection engine may receive and analyze transaction data in real-time or in batch mode. The fraud detection engine may use fuzzy logic. The fraud detection engine may use artificial intelligence such as case-based reasoning or business rules.
1. A method for using a fraud detection engine on a computing system, comprising:
at the fraud detection engine on the computing system, receiving transaction data from transactions between cardholder accounts and acceptors; and
with the fraud detection engine on the computing system, analyzing transaction data using fuzzy logic, comprising:
identifying, with fuzzy logic, transactions that have a high risk of being fraudulent, wherein identifying the transactions that have a high risk of being fraudulent comprises identifying in real-time the transactions that have a high risk of being fraudulent;
if a transaction has a high risk of being fraudulent, placing the transaction in a set of high-risk transactions to be analyzed; and
identifying sources of fraudulent transactions by analyzing the transactions in the set of high-risk transactions, wherein identifying the sources of fraudulent transactions comprises identifying the sources of fraudulent transactions in batch mode and wherein the computing system has a distributed architecture.
2. The method defined in claim 1 , wherein identifying the sources of fraudulent transactions comprises:
identifying phony acceptors, wherein phony acceptors comprise acceptors that have a high rate of fraudulent transactions.
3. The method defined in claim 2 , wherein identifying the sources of fraudulent transactions further comprises:
identifying compromising acceptors, wherein cardholder accounts that have had transactions with compromising acceptors have a higher rate of future fraudulent transactions.
4. The method defined in claim 1 , wherein analyzing the transaction data using fuzzy logic further comprises analyzing transaction data in real-time using fuzzy logic.
5. The method defined in claim 1 , wherein analyzing the transaction data using fuzzy logic further comprises analyzing transaction data using case-based reasoning.
6. The method defined in claim 1 , wherein analyzing the transaction data using fuzzy logic further comprises analyzing transaction data in batch mode using fuzzy logic.
7. The method defined in claim 1 , wherein identifying the transactions that have a high risk of being fraudulent comprises:
determining each transaction's risk of being a fraudulent transaction; and
assigning a fuzzy value to each transaction that indicates the transaction's risk of being a fraudulent transaction.
8. The method defined in claim 1 , wherein identifying the sources of fraudulent transactions comprises:
determining each acceptor's risk of being a phony acceptor, wherein phony acceptors comprise acceptors that have a high rate of fraudulent transactions; and
assigning a fuzzy value to each acceptor that indicates the acceptor's likelihood of being a phony acceptor.
9. The method defined in claim 1 , wherein identifying the sources of fraudulent transactions comprises:
determining each acceptor's risk of being a compromising acceptor, wherein cardholder accounts that have had transactions with compromising acceptors have a higher rate of future fraudulent transactions; and
assigning a fuzzy value to each acceptor that indicates the acceptor's likelihood of being a compromising acceptor.
10. The method defined in claim 1 , wherein receiving the transaction data from transactions between cardholder accounts and acceptors comprises receiving transaction data from authorized and unauthorized transactions.
11. The method defined in claim 1 , wherein identifying the sources of fraudulent transactions comprises:
identifying acceptors having transaction amounts that are unusually high as compared to transactions amounts of other acceptors, wherein the transaction amounts were accumulated following a possible fraudulent transaction.
12. The method defined in claim 1 , wherein identifying the sources of fraudulent transactions comprises:
identifying acceptors that have transaction amounts that are unusually high in a given time period as compared to transaction amounts for the same acceptors in earlier time periods.
13. The method defined in claim 1 , further comprising:
assessing transaction authorization requests received from corresponding acceptors in real-time;
comparing the corresponding acceptors with a black list of suspicious acceptors; and
denying the transaction requests received from corresponding acceptors that appear on the black list of acceptors.
14. The method defined in claim 13 , wherein the authorization requests have corresponding cardholder accounts, the method further comprising:
comparing the corresponding cardholder accounts with a list of high-risk cardholder accounts; and
denying the authorization requests that correspond to cardholder accounts that appear on the list of high-risk cardholder accounts.
15. The method defined in claim 13 , wherein the authorization requests have corresponding cardholder accounts, the method further comprising:
comparing the corresponding cardholder accounts with a list of high-risk cardholder accounts; and
adjusting risk scores of the authorization requests that correspond to cardholder accounts that appear on the list of high-risk cardholder accounts.
16. The method defined in claim 13 , wherein the authorization requests have corresponding cardholder accounts, the method further comprising:
comparing the corresponding cardholder accounts with a list of high-risk cardholder accounts; and
increasing risk scores of the authorization requests that correspond to cardholder accounts that appear on the list of high-risk cardholder accounts.
17. A method for using a fraud detection engine on a computing system, comprising:
at the fraud detection engine on the computing system, receiving authorization requests between an acceptor and an acquirer;
with the fraud detection engine on the computing system, analyzing authorization requests using fuzzy logic, comprising:
identifying, in real-time, fraudulent authorization requests;
placing the fraudulent authorization requests in a set of high-risk authorization requests; and
analyzing, in batch mode, the fraudulent authorization requests in the set of high-risk authorization requests to identify sources of fraudulent authorization requests, wherein the computing system has a distributed architecture;
identifying authorization requests received from acceptors on a black list of acceptors; and
placing cardholder accounts associated with authorization requests received from the acceptors on the black list of acceptors on a list of high-risk cardholder accounts.