The Development and Rapid Evaluation of the Knowledge Model of ADVOKATE: an Advisory System to Assess the Credibility of Eyewitness Testimony

ADVOKATE, a web-enabled knowledge base application operating in the forensic and legal domain, offers assistance to police and legal counsel in their task of assessing the credibility of potential witnesses. The knowledge model of ADVOKATE is presented in two parts. Directed graph techniques are used to model rule based knowledge and discretionary decisions and argumentation are modeled using a technique derived from Toulmin argumentation. The knowledge was acquired from legal cases, published legal and psychological research and interviews with domain experts. The model was formatively and rapidly evaluated by using it as the basis for a prototype web deployed application built using the expert system shell, WebShell. ADVOKATE, implemented as a browser accessible application, was made available to forensic experts, lawyers and police who provided feedback to the designers. The knowledge model was iteratively refined and enhanced and is available at http://advokate.bromby.vze.com/. The paper concludes with a discussion of some knowledge modeling issues demonstrated in the ADVOKATE project and the benefits provided by using a prototyping methodology to conduct a formatively biased rapid evaluation of a knowledge model in projects with limited resources.