Towards a Hybrid Human-Machine Discovery of Complex Movement Patterns

Results of automated detection of complex patterns in temporal data, such as trajectories of moving objects, may be not good enough due to the use of strict pattern specifications derived from imprecise domain concepts. To address this challenge, we propose a novel visual analytics approach that combines expert knowledge and automated pattern detection results to construct features that effectively distinguish patterns of interest from other types of behaviour. These features are then used to create interactive visualisations enabling a human analyst to generate labelled examples for building a feature-based pattern classifier. We evaluate our approach through a case study focused on detecting trawling activities in fishing vessel trajectories, demonstrating significant improvements in pattern recognition by leveraging domain knowledge and incorporating human reasoning and feedback. Our contribution is a novel framework that integrates human expertise and analytical reasoning with ML or AI techniques, advancing the field of data analytics.