Investigating suspicious vessel behaviour in light of context

Abstract

Hybrid threat events are rare and cannot be modelled solely based on data. Instead they require a focus on discovery of emergent knowledge through information sharing across agencies and systems. However, multi-intelligence can bring about reasoning challenges with multiple sources such as confirmation biases. In this paper, we present how context can be used to combat these reasoning biases. Firstly, we show how it can reduce the impact of the overly confident sources and secondly, how it can be used to provide counter-evidence. It is shown that when context is used in such a manner the reasoning results display less false confidence while still supporting the original hypothesis. We apply the reasoning scheme to the post-analysis of a real case event. The story of Andromeda was widely reported upon when the vessel loaded with 410 tonnes of explosives supposedly sailing to Libya was arrested near Crete in early 2018. Using media headlines, AIS signals and analyst reports, we show how realistic, uncertain, heterogeneous reports and contextual information can be put together to reason about its intent. We propose a reasoning model framed within the theory of evidence to combine the information from these sources. The modularity of our method allows us to easily compare different approaches to context-aware reasoning. We finally conclude on future steps for this work.