Here is the software that manages to unmask the liars – The Gazzettino

Antonio Bonanata

Here is the software able to unmask the lies. They are full of lies (and, with them, little secrets, hidden truths, actually concealed) virtually all human relationships: friendships to the parent-child bond; from professional relationships with those that are established in judicial inquiries. And perhaps it would be just that the scope of most use to derive the maximum benefit, the software developed by some computer University of Michigan, who worked on a project designed by the Department of scientific research in the Pentagon. The main innovation of this software lies in its autonomy: all the systems used for interrogation, the famous “lie detectors” to the Facial action coding system (in other words, the method of analysis of the face ) were based on the necessary human input, reading the data to interpret and give a result.

There are many reference parameters: heart rate, breathing, blood pressure, even the most subtle involuntary movements of the face; all contributed to unmask the liar turn, perhaps under pressure from a shrewd investigator but adept at cover under a veil of imperturbability their emotions. Well, just now that this new software steps through a sieve videos of testimony and interrogation, without the help of an expert, to flush out any lies. Computer scientists who have programmed explain that the success rate may reach 75 percent of cases, compared to 53 percent of the systems which make use of human collaboration.

“The difference is explained by the difficulties for investigators to record all the minute clues “illustrates Rada Mihalcea, project manager. But at what “training” was performed on the software set up the University of Michigan to be so effective? The team has done to analyze the system the filming of 120 processes and studying the statements of witnesses and defendants, every gesture, every word, every subtle inflection has been stored; consequently, the software has learned to connect the repetitiveness of the gestural sample the outcome of the process. In this way, the system is able to statistically evaluate how they move, speak and as such expressions are liars. Progressing well, enabling the construction of a virtual archive, to date, that investigators and prosecutors can refer to associate, for example, a certain facial movement signal of truthfulness (or falsehood) of a certain individual.

Already we can say, however, what the “typical characteristics” of serial liar: grimace or frown are recorded in 30 percent of cases (and only in the 10, when it is telling the truth). To face those who are questioning, strange to say, is more frequent in cases of lies than in sincerity (70 percent vs. 60). Gesturing with both hands is a sign of lying in 40 percent of cases (but, if you say the truth, the percentage drops to 25). More predictable, associating the use of the third person – instead of the first – to a little true story, the removal rate from what you are saying. Not to mention the interjections and all those manifestations of indecision, as “er …” or “uhm …”, with which they try to fill the empty space of silence. In short, for liars, they are partners committed to hide a treason or criminal intent to construct an alibi, promises hard life.
     
     
          

© ALL RIGHTS RESERVED