Arguing about uncertain heterogeneous information for threat assessment

This chapter presents a mixed initiative threat assessment model combining evidential and abductive reasoning. The evidential reasoning produces pro and contra arguments for the each threat component (intent, opportunity, and capability) while considering uncertain heterogeneous (soft and hard) transient information as pieces of alternative stories (hypotheses) based on "what might happen." The belief-based abductive reasoning computes probabilities of these stories by combining beliefs in all the arguments expressed within different uncertainty frameworks (probability, possibility, beliefs). Beliefs in the arguments are fused and used for selecting decisions on either alerting an analyst on potential or imminent threat, or as a basis for continuing observations for making a better decision. The fusion process is complicated by different models used for representing uncertainty characterizing soft and hard data. The chapter addresses this problem by utilizing a unified framework of credibility representation within the Transferable Belief Model. The results of a case study designed to illustrate the described model are also presented.