Visual Data Mining: A Great Opportunity for Criminal Investigation

Introduction

Current generation criminal justice relies mostly on manual procedures and processes which are time-consuming, hard to scale, error-prone and sometimes invasive. For example, the polygraph commonly referred to as the “lie detector test” is based on a process that measures and records several physiological indices such as blood pressure, pulse, respiration, and skin conductivity while the subject is asked and answers a series of questions. The expectation is that deceptive answers will produce physiological responses that can be differentiated from those associated with non-deceptive answers. A polygraph test consists of only “yes” or “no” questions which are dependent on the ways in which several physiological activities are simultaneously recorded during a test and aims to monitor at least three bodily reactions to determine whether a person is truthfully answering questions. These reactions include the respiratory rate, sweat gland activity, and cardiovascular activity. The polygraph test has been at the heart of debates between proponents and opponents of the system, some claiming a 90% success rate while others questioning its accuracy due to many factors, including 1) not being reliable and unable to tell the examiner whether the subject is telling the truth or lying as they depend on the physiological reactions to the questions asked and 2) possibly swaying the examiner as they rely on the physiology of individuals who may be very nervous, unshakable or capable of controlling their physical reactions in order to defeat the lie detection exercise.

The criminal justice of the future is expected to be empowered by the most modern information and communication technologies (ICTs) to provide magistrates, advocates, judges, members of the jury and the police force a rich set of services such as virtual court presence and hearing participation through visual sensor networks (VSNs). These networks will be equipped with camera sensors and communication capabilities to enable recording and streaming to places where virtual participation is required. In addition, machine learning techniques and statistical methods can be leveraged in order to detect and provide a rich set of indicators to these participants, including lip-shape and facial micro-expressions, among others. In this regard, facial micro-expressions have been shown to be strong indicators of deception, giving rise to the possibility of detecting deception automatically. This chapter revisits the issue of deception detection by proposing visual data mining (VDM) as a non-invasive alternative to deception detection in next generation criminal justice.