Detection of special scams by emotion analysis

Abstract

Telephone scams called "special scams" have become a major problem, with people being defrauded of nearly 30 billion yen every year in Japan. In this study, we used emotion analysis techniques based on speech analysis to detect whether victims were being deceived by analyzing multiple emotional parameters of victims in a special scam. In the verification, we first clarified the characteristics of the emotions generated in victims by two scam methods. Next, we improved the accuracy of detecting the risk of a victim being deceived by limiting the analysis section.