You are here:  Home  >  Data Education  >  Current Article

Machine Learning Part Of New Versium Solution To Quickly Find Fraudsters

By   /  February 19, 2014  /  No Comments

versiumby Jennifer Zaino

Machine learning is playing a role in fraud prevention: This week Versium launched its Predictive FraudScore solution to help companies weed out fraudsters from signing up for their services or conducting ecommerce transactions with them. All the organization needs is an email address.

The solution is based on Versium’s LifeData predictive analytics platform that also is behind the company’s churn, social influencer, shopper and custom scoring products. “There are three fundamental areas we bring to fraud scoring: unique data, powerful matching technology to identify and associate that data to accounts or consumers as they sign up, and applying machine learning to that unique data set to predict whether that account is likely to be associated with fraud or not,” says Versium’s CEO Chris Matty.

The FraudScore service provides an enterprise a very strong indication of whether a person is legitimately interfacing with it at the point in time that that entity registers with the company. “That’s quite upstream from where normal fraud prediction takes place,” Matty says.

The solution works in cases such as account hijacks, where someone takes over someone else’s credentials but changes information such as the email address for confirming product orders or the physical address to which a product would be delivered, since they don’t want the real identity holder to learn about the transaction or receive the goods. It’s becoming increasingly difficult for companies to suss out consumer identity thefts with traditional authentication and prevention efforts as fraudsters become more sophisticated and as enterprises themselves lack first-hand knowledge of applicants, the company says. “Traditional rules-based systems are being overcome based on the sophistication of fraudsters,” Matty says.

Versium’s LifeData platform, however, has 4 billion email addresses aggregated over time, which it can tap into along with on- and off-line data characteristics that define who consumers are. “We look at categorically about 30 or 40 different data signals,” Matty says, including online presence and behavior indicators and offline attributes such as demographic data, purchase interest data, and more from publicly or commercially available sources. Its patent pending technology assigns and associates data matches, and it applies machine learning and predictive analytics on top of what totals up to 300 billion LifeData attributes to output a predictive FraudScore that’s delivered as a service. Higher scores warrant a closer look, lower scores are probably safe bets. It reports fraud capture rates as high as 85 percent.

“We index starting with a piece of information tied to the individual and look for presence via all different mechanisms,” Matty says. “Those committing fraud are creating new identities or taking existing real ones and using then in inappropriate ways, and based on the association of this data we can determine whether an identity is created for the purpose of fraud or if computer modeling can say that the person is not doing fraudulent things.” For instance, a different address applied to a credit card is a piece of data the solution would look at for fraud signals.

Indeed, Matty points out, those looking to commit fraud also are not using their own normal, personal data presence to engage in those activities, since they don’t want to get caught, and that very dearth of information that can be tied to them may be a signal of fraudulent intent, he says. “Not finding a piece of data in applying computer modeling is a signal in itself” he says. “That is data.”



About the author

Jennifer Zaino is a New York-based freelance writer specializing in business and technology journalism. She has been an executive editor at leading technology publications, including InformationWeek, where she spearheaded an award-winning news section, and Network Computing, where she helped develop online content strategies including review exclusives and analyst reports. Her freelance credentials include being a regular contributor of original content to The Semantic Web Blog; acting as a contributing writer to RFID Journal; and serving as executive editor at the Smart Architect Smart Enterprise Exchange group. Her work also has appeared in publications and on web sites including EdTech (K-12 and Higher Ed), Ingram Micro Channel Advisor, The CMO Site, and Federal Computer Week.

You might also like...

Property Graphs: The Swiss Army Knife of Data Modeling

Read More →