Deception Detection Using Human Reasoning
Author | : Deqing Li |
Publisher | : |
Total Pages | : 338 |
Release | : 2013 |
ISBN-10 | : OCLC:855857534 |
ISBN-13 | : |
Rating | : 4/5 ( Downloads) |
Download or read book Deception Detection Using Human Reasoning written by Deqing Li and published by . This book was released on 2013 with total page 338 pages. Available in PDF, EPUB and Kindle. Book excerpt: Scientific methods that detect deception have been studied since 1895. Among them, computational methods have gained popularity in recent decades thanks to the development of artificial intelligence, (AI). Detecting deception by categorizing verbal/non-verbal cues using machine learning techniques has been the main stream approach in the field. We investigate deception detection methods that work on communication content in a written format. In this work, we propose that verbal/non-verbal cues are simply artifacts created during the implementation of deception, and we instead study the cognitive process behind the formation of deception. We detect deceptive communication by modeling the cognitive process in deception, comparing the semantic structure of deceptive communication with that of honest communication, and identifying the patterns for deceptive reasoning. Our method differs from existing works by targeting at malicious intent instead of wrong information, by deriving observations directly from the intent to deceive, and by taking individual difference into consideration. As a result we are able to distinguish unintentional misinformation from intentional deception, an approach which no existing research has yet addressed. In representing the reasoning process of human communication we use Bayesian Networks. The contributions of our work lie with (i) its development of an alternative method of deception detection and improvement of detection performance by using the cognitive process in human argumentation, (ii) its exploration of the deep cognitive process in human argumentaion through linguistic information, (iii) its ability to explain the way that a deceptive communication is formed and detected, (iv) its intuitive representation of deceptive reasoning, which facilitates the corresponding explanation of the verbal cues of deception, and (v) its analysis of the impact of different types of deception datasets on the detection performance. We propose to compare our approach with verbal cues in terms of accuracy and reliability. Our ultimate goal is to obtain a better understanding of how humans reason, the decisions they make when they decide to deceive.