Deception Detection

Principal Investigator: Victoria Rubin

Research Assistants: Niall Conroy, Tatiana Vashchilko, Yimin Chen (and previously, Jeff Taylor, Sarah Camm, and Jason Neal)

Deception is pervasive in computer-mediated communication technologies such as e-mails, weblogs, or message boards. It takes various forms: falsifying, omitting material facts, and dodging issues by changing the subject or by offering indirect responses.

On average, people succeed 54% of the time in discriminating deceptive messages from truthful ones. Human abilities to detect deception can be enhanced with automated tools that identify and summarize objective differences in verbal cues that occur when people lie, compared to when they tell the truth. Current tools achieve 74% success rates but are focused on a general sense of deceitfulness. This research program sets out to develop an algorithm for an automated classification system, AutoLieDetect, that is capable of discerning deceitful messages and classifying them by their suggested variety (such as an exaggerated claim) and their potential deceitful content (such as a person's quality).

In addition to an experimentally collected truthful and deceptive messages of various types, a large-scale dataset of people's beliefs, discussions, and reports about deception will be collected from web-based publicly available communication applications. Both datasets (of experimentally obtained and freely occurring retrospections) will be analyzed for reliable verbal cues statistically and linguistically, with a unique combination of state-of-the-art Natural Language Processing techniques. The identified verbal cues will become input to predictive Machine Learning algorithms, and the prediction outcomes will serve as the basis for AutoLieDetect. This decision support tool will augment human abilities to deception detection and support credibility assessment tasks in ordinary computer-mediated communication (e.g., in Facebook).

Successful implementation of such automated deception detection and classification has an enormous potential for beneficial employment in professional settings as well, such as in analysis of written complaints in mediation between consumers and businesses under consumer protection regulations, and more generally, in law enforcement and jurisprudence, and in intelligence gathering for Canadian national security.


Associated Publications:

  • Rubin V.L., Conroy N. (2012) Discerning Truth from Deception: Human Judgments and Automation Efforts. First Monday, Vol. 17, No. 3 at http://firstmonday.org

  • Rubin, V.L., Vashchilko, T. (2012) Identification of Truth and Deception in Text: Application of Vector Space Model to Rhetorical Structure Theory. In Proceeding of the 13th Conference of the European Chapter for the Association for Computational Linguistics: Computational Approached to Deception Detection Workshop, Avignon, France, April 23, 2012 (EACL 2012: http://eacl2012.org/home/index.html)

  • Rubin, V.L., Conroy, N.J. (2011). Challenges in Automated Deception Detection in Computer-Mediated Communication. at the 74th Annual Meeting of theAmerican Society for Information Science and Technology: Bridging the Gulf: Communication and Information in Society, Technology, and Work, New Orleans, Louisiana, October 9-12, 2011. ASIS&T2011

  • Rubin, V.L., Camm, S.C. (Under Review). Deception in Video Games: Examining Varieties of Griefing. The Online Information Review. Through Emerald

  • Rubin, V.L., Camm, S. C., (2011) Griefing and Deception in Video Games: Examining Attitudes towards the Phenomena. The 39th Annual Conference of the Canadian Association for Information Science: Exploring Interactions of People, Places and Information (CAIS / ACSI 2011), Fredericton, N.B., Canada, June 2 - 4, 2011

  • Rubin, V. L. (2010). On Deception and Deception Detection: Content Analysis of Computer-Mediated Stated Beliefs. Proceedings of the American Society for Information Science and Technology Annual Meeting: Navigating Streams in an Information Ecosystem,October 22–27, 2010, Pittsburgh, PA, USA. ASIS&T 2010

  • Rubin, V. L., & Liddy, E. D. (2006). Assessing Credibility of Weblogs. Proceedings of the AAAI Spring Symposium: Computational Approaches to Analyzing Weblogs (CAAW), 27-29 March, 2006, Stanford University. AAAI:CAAW, 2006

Back to top



LITRL Logo
only search this site
Copyright © 2011-2012, Victoria Locktionova Rubin. All rights reserved.
Last Updated: