Abstract

It has long been hypothesized that the ability to deceive depends on personality - some personality types are `better' at deceiving in that their deception is harder to recognize. In this work, we evaluate how the pattern of personality of a speaker affects the effectiveness of machine learning models for deception detection in transcripts of oral speech. We trained models to classify as deceptive or not deceptive statements issued in Court by Italian speakers. We then used a system for automatic personality recognition to generate hypotheses about the personality of these speakers, and we clustered the subjects on the basis of their personality traits. It turned out that deception detection models perform differently depending on the patterns of personality traits which characterize the speakers. This suggests that speakers who show certain types of personality also have a communication style in which deception can be detected more, or less, easily.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call