Automating the Application of Deception Detection Heuristics to Open Sources

Period of Performance: 09/08/2003 - 12/08/2005

$8.9K

Phase 2 SBIR

Recipient Firm

Stottler Henke Associates
1650 South Amphlett Boulevard, Suite 300
San Mateo, CA 94402
Principal Investigator

Abstract

We propose to construct an intelligent software agent system for identifying misleading information, called Skeptic, which exploits scaffolding provided by a collection of largely domain-independent deception detection heuristics. These heuristics provide Skeptic with a significant advantage over purely inductive methods by allowing it to exploit the adversarial nature of this problem. The result is a system that will effectively anticipate actions of the adversary and inform an evidential reasoning system. Skeptic will be the first deception detection system to employ information extraction to allow the detection of misleading information from Web sites and other open sources. Further, Skeptic will adapt over time, which means it can be deployed early, and mature as the understanding of the problem matures. In our Phase I investigation, which targeted a particular class of deception called "stock pumping," we were able to verify the utility of goal-directed evidence collection from very diverse evidence sources. We also discovered the value of employing heuristic means for identifying new targets for investigation, enabling simultaneous classification of instances of suspected stock pumping and identification of new instances. Our Phase II effort will result in a fully functioning prototype of Skeptic with applicability to a variety of homeland defense and commercial applications.