Soundwave
Decepticons Research Group

Digital Deception in Computer Mediated Communication

Welcome!

This is the home of Lie-M, a data collection resource for Cornell's Computer-Mediated Communication Laboratory. "The Decepticons" (as the deception research team calls itself) study deception among psychological, social, and cultural dimensions. In particular, the group is interested in studying how technology affects digital deception. Digital deception refers to any intentional control of information in a technologically mediated message to create a false belief in the receiver of the message.

Research Questions

The Decepticons' research is motivated by the following questions.

      Deception Production
    • Do people lie most frequently face-to-face, on the telephone, or by email?
    • Do people use different media to lie about different types of things, or to different types of people?
      Deception Detection
    • Are we worse at detecting a lie in a text-based interaction than we are in face-to-face?
      Deception Analysis
    • Can automated techniques be developed to analyze the linguistic properties of deceptive and truthful messages?
    • Can automated techniques be developed to analyze the linguistic properties of deceptive and truthful messages?
    • Can applications based on these approaches be developed to assist in real-time deception detection in online communication?

Lie-M and the Lie-brary

Lie-M is a real-time instant messaging (IM) client for self-reported deception classification. Lie-M is modified from Gaim, an open-source instant messaging client that connects to popular IM protocols, and is licensed by the LGPL.

Lie-M asks users to classify their messages before they send them to their intended recipients. If the message is deceptive, the user is asked to rate how deceptive it is on a scale from 1-5 (slight to very deceptive) using a small group of radio buttons built into the IM window. This message, along with its rating, is sent to our server for analysis. The messages of the conversation partner are not sent to our server, nor is the deceptive rating sent to this person. The message is devoid of personally identifying information (such as screen names).

Developed by Tucker Barrett, Lie-M is and will be used to collect self-reported measures of deception on a massive (500,000-1,000,000 messages) scale. The resulting corpus, the Lie-brary, will be used to study linguistic and temporal correlates of deception. An example of a temporal investigation will involve measuring how often people typically lie online, and a linguistic investigation will example what types of words or syntax patterns are common.

This is the official webpage (developed for BOOM 2005) of the Lie-M project, which will be presented by technical lead, Tucker Barrett.