Capturing consensus knowledge from multiple experts

The acquisition of knowledge from multiple experts for the development of intelligent decision support systems is difficult in that divergent opinions are often held so strongly that consensus cannot be reached. Yet, disagreement is a common feature in fields such as law, science and engineering. A model for eliciting common ground amongst experts without the involvement of extensive knowledge engineer resources is presented. The model, called Consult, enables experts to elicit a set of arguments that are plausible for an issue. The process involves three steps. A knowledge engineer initially elicits a set of arguments consistent with the generic actual argument model (GAAM) from one expert. A process based on the Delphi method is then used to anonymously elicit suggested modifications from other experts. The Borda Count voting system is used in the third phase to elect modifications preferred by experts. In this way, Consult semi-automates knowledge acquisition by capturing a consensus view from multiple experts.