Browse Prior Art Database

Smarter ATM analytics for fraud detection

IP.com Disclosure Number: IPCOM000245552D
Publication Date: 2016-Mar-16
Document File: 3 page(s) / 172K

Publishing Venue

The IP.com Prior Art Database

Abstract

Briefly, the core idea of our invention is to score in real time, each ATM transaction based on customer’s behavior (data mining on historical structured data) and many new sources, unstructured data such as twitter information (geo localized , close to ATM device), weather data, time (when customer requesting ATM operation), and customer mobile phone data. At the end, once the PIN code has been validated (normal ATM check process remaining the same process – still need to validate PIN), back end office ATM solution scores in real time each new data sources, and gives a final score (between 0 and 10 – 0 means 100% sure is not a fraud, 10 means 100% sure is a fraud) ,granting or not the ATM operation with our invention : 1- you can be located nearby the ATM (twitter is locating you close to the ATM) but under constraints (with gun for instance) , we could have detected this situation due to many tweets talking about "gun" "aggression" "feeling about insecurity" 2-With our invention, we add many other sources (social data, weather, mobile) to improve scoring algorithms 3-thief could have theft your credit card and your mobile phone. But if you are 90 years old lady and it is snowing, at 2AM , our algorithm will refuse the ATM transaction, even if the lady's mobile phone is present near by the ATM 4-we do not need to add any additional user action or any additional devices from ATM side ... 5-with our idea, we create scoring method to detect ATM fraud, scoring multiple sources of information integrateing unstructured data such as social media data or weather data, time data. This invention creates scoring method more accurate and based on more data sources than this predictive model based on variables

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 01 of 3

Smarter ATM analytics for fraud detection

Fraudulent use of automated teller machine (ATM) systems has become a substantial problem for banks and other financial institutions. 2.75 Billions $ of fraud per year, where Customer complaints have been received that "phantom

withdrawals" have been made from their accounts by persons passing themselves off as the customers, and bank increase the cost of "credit card" to manage this fraud.

Indeed, as of today, if someone stole your bag, with "credit card" with your

Personal Identifier Number (PIN) code (written in the bag), he can enter your PIN and get 1000$ from your account. Or the same situation where you were asked to provide PIN code under pressure/aggression.

Problem : PIN Code is the only one security check during ATM withdrawal process.

    Briefly, the core idea of our invention is to score in real time, each ATM transaction based on customer's behavior (data mining on historical structured data) and many new sources, unstructured data such as twitter information (geo localized , close to ATM device), weather data, time (when customer requesting ATM

operation), and customer mobile phone data.

    At the end, once the PIN code has been validated (normal ATM check process remaining the same process - still need to validate PIN), back end office ATM solution scores in real time each new data sources, and gives a final score (between 0 and 10 - 0 means 100% sure is not a fraud, 10 means 100% sure is a fraud) ,granting or...