On a text-processing approach to facilitating autonomous deception detection

Current techniques towards information security have limited capabilities to detect and counter attacks that involve different kinds of masquerade and spread of misinformation executed over long time periods to achieve malicious goals. Detection of such deceptive information obtained during online interactions (emails, chat room conversations) is the first step before counter strategies can be developed. With the large-scale use of information technologies as a general communication medium, facilitating deception detection is a key enabler to utilizing information systems to their fullest potential. This article presents a framework for computer-aided deception detection building on the interpersonal deception theory (IDT) of human interpersonal communication research and text-processing techniques for facilitating deception analysis of text-oriented communication. A state-transition diagram based framework is proposed to model the dynamic evolution of an interpersonal conversation between a sender and receiver based on the IDT-based process schemata. The framework is then utilized to develop a deception detection agent to process textual information. Deception detection is defined as a process of model verification. Architecture of a prototype under development and open problems for further research in this area are outlined.