Supporting the intelligence analysis stages with approximate reasoning: methods and tools based on granular computing

Abstract

Current threats, such as terrorism and cyber-terrorism, pose new challenges to security and defence communities, and the ability to reason with different perspectives and detecting connections between facts, relationships and events becomes crucial to address these challenges. To this purpose, a less procedural and standardized approach is useful, able to leverage current computational and artificial intelligence technologies to detect threats and protect physical and cyber-physical systems. This raises a strong interest in defining and adopting new methods and techniques that, to a certain extent, are such as to replicate, or at least to support, human cognitive processes. If we consider the "creativity" behind some recent attacks, such as that one of 09/11/2001, we can understand the need to rethink security in situational rather than procedural terms, and this has an important implication that helps to frame the problem and research objectives of this thesis. This implication is a shift from being aware of what we need to prevent (and the related rules and procedures to that end) to gaining greater awareness of what might happen. From this consideration, it emerges the need of methods and tools that support decision makers in their ability to carry out analyses that allow to hypothesize different threat scenarios, and to reason about their evolutions. This is, essentially, the main objective of the so-called intelligence activities. The research problem investigated in the Ph.D period is how to improve the awareness of analysts and decision makers in the early stages of an intelligence analysis to prevent intentional attacks, and the specific objectives concern the definition and validation of reasoning methods based on Granular Computing (GrC) for this purpose. ... [edited by Author]