How machine learning helps prevent online daters from getting scammed

The number of people looking for romance online is increasing, and so is the presence of scammers on dating sites and apps. Here's how machine learning is being used to catch them.

How machine learning is used to defend networks and catch hackers

Ahead of Valentine's Day 2018, CBS News reported that the FBI is warning those looking for love online to be wary of scammers on dating sites and apps. Online dating services can be lucrative targets for fraudsters. According to the article, online dating scams cost Americans up to $210 million over the past three years.

SEE: Every three hours, an online dater gets scammed (CNET)

According to Kevin Lee, architect at fraud prevention and machine learning firm Sift Science, a typical online dating scam goes like this:

"A fraud scammer sets up a series of false personal profiles on these online dating sites, and then reaches out to form relationships. At first, the relationship seems pretty genuine. Then, once the relationship is formed, at some point the scammer determines an 'extraction point' where she or he begins to ask the other party for money to 'help them out.'"

"We've reached a point where one in three new couples now meet online, instead of at the traditional bar or at parties," said Lee. "Unfortunately, with the increase of online dating, online dating fraud has also accelerated."

According to Lee, the fraud problem may be much bigger than any of the reported numbers: "Because usually only 10% of people report these incidents, since many are too embarrassed to do so."

Lee's company is working on a solution to the problem that incorporates machine learning and analytics. "Analytics can extrapolate the percentage of online dating accounts that are fraudulent versus legitimate, and they can also identify the accounts that have a high probability of being fraudulent," he said.

A combination of analytics and machine language identifies users who have multiple profiles on a dating site by tracing the originating points of the accounts to specific IP addresses, devices, etc. that are part of a user's network. "We can tie all of this activity into a single profile to identify persons who are setting up multiple accounts, and who might be engaging in fraudulent activity," said Lee.

SEE: Match Group's Tinder emerges as cash cow courtesy of data science, dynamic pricing, paying off tech debt (ZDNet)

Content can also be analyzed for language misspellings and usage errors or requests for money, which indicate scams.

"We grade each profile for fraudulent potential on a scale from 0 to 100," said Lee. "A rating of 100 would suggest high probability of fraud. A rating of 0 would indicate very low probability."

Sift Science's clients are online dating sites--but no one site uses the analytics in the same way.

"Some sites are geared toward helping users find permanent relationships, while others are more social or casual sites," said Lee. Consequently, some sites might opt to send an alert out to a user about a particular account if there is suspicion of fraud--but others might not.

"In some respects, an online dating site is less concerned about fraud than a financial institution because the immediate damage to the financial company is in dollars and cents," said Lee. "For an online dating company, there is no immediate dollars and cents loss, but the company could see brand value and reputation loss over time that will ultimately affect the bottom line."

So how can you ensure that your online date is for real?

"If you want to meet the person, but he or she repeatedly tells you that they can't, it could be a sign of fraud," said Lee. "If they tell you they are stationed abroad and in the service, many online dating sites now verify this status and issue a "badge" that is on the individual's profile that shows that the individual is in the service. You can also use resources like Google to research whether the person is who they say they are."

SEE: Lunch and learn: Dealing with the risks of identity theft (Tech Pro Research)

It also pays to be vigilant.

"Don't get so excited about who first messages you, and instead be a little cautious," advises Lee. "The best news of all for online dating users is that although fraud is there, the majority of online dating relationships are genuine, and online dating offers a fresh venue for people who want to meet others without having to go to a brick and mortar social forum."

One more tip for business leaders: When you have connected with someone, be careful what you share. TechRepublic's Olivia Krauth recently reported that a significant number of professionals are guilty of signing into dating apps from work devices, or oversharing business information and putting their company at risk.

Also see

Image: iStock/solar22

By Mary Shacklett

Mary E. Shacklett is president of Transworld Data, a technology research and market development firm. Prior to founding the company, Mary was Senior Vice President of Marketing and Technology at TCCU, Inc., a financial services firm; Vice President o...