Methods and apparatuses for fraud handling

Patent metadata

Field	Value
Publication number	US-11127015-B2
Application number	US-201916361257-A
Country	US
Kind code	B2
Filing date	Mar 22, 2019
Priority date	Mar 26, 2018
Publication date	Sep 21, 2021
Grant date	Sep 21, 2021

Abstract

The present disclosure relates to a concept of fraud handling. A data transaction request is received via a data network from at least one user account. The data transaction request is analyzed based on predefined fraud detection rules to generate a fraud score associated with the at least one user account. The at least one user account is classified as fraudulent account if the associated fraud score exceeds a predefined fraud likelihood threshold. Data transactions associated with a classified fraudulent account via the data network are pseudo-randomly permitted or blocked.

First claim

The invention claimed is:

1. A method of fraud handling, the method comprising:
training a machine learning model based on past fraud behaviors;
configuring the trained machine learning model to recognize fraudulent data transactions;
receiving, via a data network, a data transaction request from at least one user account;
analyzing the data transaction request based on predefined fraud detection rules to generate a fraud score associated with the at least one user account, wherein analyzing the data transaction request includes
feeding the data transaction request from the at least one user account into the trained machine learning model configured to recognize fraudulent data transactions;

classifying the at least one user account as a fraudulent account in response to the associated fraud score exceeding a predefined fraud likelihood threshold;
pseudo-randomly permitting or blocking data transactions associated with the classified fraudulent account via the data network, wherein a pseudo-randomly permitted data transaction is a data transaction that is permitted instead of blocked even though the data transaction is associated with the classified fraudulent account, wherein pseudo-randomly permitting or blocking data transactions is configured to intentionally obfuscate behavior of the fraud handling; and

deteriorating the permitted data transaction associated with the classified fraudulent account with respect to data transactions associated with a classified non-fraudulent account, wherein deteriorating the permitted data transactions comprises at least one of purposely using a higher latency for data communication or communicating altered data content.

2. The method of claim 1 , wherein pseudo-randomly permitting or blocking the data transactions comprises generating a pseudo-random permission-restriction-sequence based on a predefined pseudo-random seed.

3. The method of claim 1 , wherein deteriorating the permitted data transactions comprises at least one of pretending service unavailability and providing a game environment which does not allow a puzzle to succeed.

4. The method of claim 1 , wherein pseudo-randomly permitting or restricting data transactions comprises reacting in real-time to incoming data transaction requests.

5. An apparatus for fraud handling, the apparatus comprising: a receiver configured to receive, via a data network, a data transaction request from at least one user account; a processor circuit configured to train a machine learning model based on past fraud behaviors, configure the trained machine learning model to recognize fraudulent data transactions, analyze the data transaction request based on predefined fraud detection rules to generate a fraud score associated with the at least one user account, wherein the processor circuit for analyzing the data transaction request is further configured to feed the data transaction request from the at least one user account into the trained machine learning model configured to recognize fraudulent data transactions; classify the at least one user account as a fraudulent account in response to the associated fraud score exceeding a predefined fraud likelihood threshold; pseudo-randomly permit or block data transactions associated with the classified fraudulent account via the data network, wherein a pseudo-randomly permitted data transaction is a data transaction that is permitted instead of blocked even though the data transaction is associated with the classified fraudulent account, wherein pseudo-randomly permitting or blocking data transactions is configured to intentionally obfuscate behavior of the fraud handling; and deteriorate the permitted data transaction associated with the classified fraudulent account with respect to data transactions associated with a classified non-fraudulent account, wherein deteriorating the permitted data transactions comprises at least one of purposely using a higher latency for data communication or communicating altered data content.

6. The apparatus of claim 5 , wherein the processor circuit is configured to generate a pseudo-random permission-blocking-sequence based on a predefined pseudo-random seed.

7. The apparatus of claim 5 , wherein the processor circuit is configured to analyze the data transaction request, classify the at least one user account, and pseudo-randomly permit or restrict the data transactions in real-time.

Assignees

• Sony Corp

Inventors

• Vandezande Bart
• Aylward Conor
• Traum Cristian
• Lim Erbin
• Jochems Barbara
• Elshocht Olivier

Classifications

• G06Q20/4016 Primary

  involving fraud or risk level assessment in transaction processing · CPC title

• G06Q20/407

  Cancellation of a transaction · CPC title

• G06Q20/405

  Establishing or using transaction specific rules · CPC title

• H04L51/216

  Handling conversation history, e.g. grouping of messages in sessions or threads · CPC title

• H04L51/212

  using filtering or selective blocking · CPC title