Abstract
Terrorist attacks launched by extremist groups or individuals have caused catastrophic consequences worldwide. Terrorism risk assessment therefore plays a crucial role in national and international security. Fuzzy reasoning-based terrorism risk assessment systems offer a significant potential of providing decision support in combating terrorism, where highly complex situations may be involved. However, missing expertise often presents challenges for configuring systems that can otherwise assess the likelihood and risk of possible attacks due to the availability of only sparse rule bases. Hierarchical fuzzy rule interpolation systems may be adopted in order to overcome such problems. Unfortunately, situations can become more sophisticated because certain important antecedent values may be missing, which need to be inferred from the known (or hypothesised) consequences. Initial theoretical work on backward fuzzy rule interpolation has been proposed to cope with certain underlying problems. Nevertheless, little has been done in developing and applying an integrated hierarchical bidirectional (forward/backward) fuzzy rule interpolation mechanism that is tailored to suit decision support for terrorism risk assessment. This paper presents such an integrated approach that is capable of dealing with dynamic and insufficient information in the risk assessing process. In particular, the hierarchical system implementing the proposed techniques can predict the likelihood of terrorism attacks on different segments of focused attention. It also helps identify hidden variables that may be useful during the decision support process via performing reverse inference. The results of an experimental investigation of this implemented system are represented, demonstrating the potential and efficacy of the proposed approach.