From Academia: Researchers Develop AI Scam Detection

A new study has examined the ways in which AI can be used to tackle dating service scammers and fraudsters.

The authors, representing King’s College London, University of Bristol, Boston University, and the University of Melbourne, titled the work “Automatically Dismantling Online Dating Fraud”.

They found that 60% of scammers were men, and that they had an average age of 50. The research also showed words like “passionate” and “loving” were often employed.

Training a data set on 15,000 free Dating ‘N More profiles, the researchers then looked to devise a system that would be able to spot fake accounts as they register.

It took into account factors such as age, gender, occupation, text and image data.

There was some discussion of a type 1 and 2 error problem – the more trigger-happy scam detection software is, the more likely it is to wrongly identify legitimate users as scammers. The more conservative the software is, however, the easier it will be to evade.

False positives (wrongly identifying legitimate users as scammers) occurred between 0.1% and 1.4% of the time.

The conclusion reads: “Our experimental results show that our system can accurately detect online dating fraud profiles, with high precision. A case by case analysis of our results, however, indicates that there are certain false profiles that look genuinely real. For these cases, we have noted that other sources of information, such as the messages exchanged, could be very informative.

“As future directions, we aim to more broadly examine the available data on online dating fraud, seeking information actionable for enforcement and other countermeasures.”

Find the study here.