When a person is interviewed, some of the answers may be inaccurate, or even deceptive, because the person may have either incomplete information, is telling only part of the truth, or is fabricating a false answer, or a combination of all three. When the person is habitually making statements that are known to be false, or only partly true, emotional and/or intellectual conflicts often arise within them, and these conflicts may become manifest by inconsistencies in use of different parts of speech or in logical relationships between statements. These inconsistencies are more subtle than inconsistencies in factual statements, and identification of these inconsistencies is more difficult and less straightforward than identification of factual inconsistencies.
This innovation is a computer-based method for analyzing a collection of transcribed documents and/or reporting interviews of a person. It has been developed to determine whether the person is experiencing high stress, emotional volatility, or internal conflict in their responses during the interview.
In this system, one or more of four procedures is applied: a first statistical analysis, a second statistical analysis, a third analysis, and also a heat map analysis to identify one or more documents containing the person’s responses. Words in the documents are characterized in terms of dimensions representing different classes of emotions and states of mind, in which the person’s responses that manifest high stress, emotional volatility, and/or internal conflict are identified. A heat map visually displays the dimensions manifested by the person’s responses in different colors, textures, geometric shapes, or other visually distinguishable indicia. The person’s use of particular language, including colloquialisms and word order in the person’s asserted home region, and knowledge of an asserted line of work, are also analyzed.
The person’s responses are analyzed by reference to different dimensions, with each dimension having a collection of words and phrases relating to an emotion or state of mind. A subset of dimensions is identified that is more prominent in the documents containing the person’s responses. This subset of dimensions is examined more closely to identify responses in which the person manifests the indicators of deception previously enumerated.
This work was done by Charles Jorgensen of Ames Research Center and Rajkumar Thirumalainambi of Perot Systems. Additional information is available at http://www.nasa.gov/ames-partnerships/technology/detect-stress .
Inquiries concerning rights for the commercial use of this invention should be addressed to
Refer to ARC-16407-1.