|эдак можно и расширения алгоритма удумать :(|
Researchers from City University in London detailed their findings in a paper set to be published in Journal of Management Information Systems. The paper, titled "Untangling a Web of Lies: Exploring Automated Detection of Deception in Computer-Mediated Communication," describes tell-tale signs that someone is lying.
Deceitful emailers have a tendency to avoid personal pronouns—such as “I” and “me”—while also including language that flatters the recipient. The paper’s authors say their findings could be used within a business context, to identify corporate deception and spot when a customer might not be telling the truth.
“Everybody lies and most companies realize that the customer is not always right,” said Ko de Ruyter, professor of marketing at Cass Business School, City University. “In fact, customers can often be dishonest and it is costing companies a lot of money. Our lie detection software can help companies to assess whether their customers bend the truth in their favor and to decide whether they want to continue doing business with them.”
The algorithm proved to be a better judge of lies than the average human, successfully detecting lies 70 percent of the time compared to the 54 percent success rate of humans.
“Humans are startlingly bad at consciously detecting deception,” said Tom van Laer, a senior lecturer in marketing at Cass Business School.
“This research opens up the possibility of fraud prevention and deception detection technology across lots of in-person domains, not just email.”