Mining contacts from spatio-temporal trajectories

Contact mining is discovering objects inclose proximity in their movements in order to reveal possible interactions, infections, collisions or contacts. This process can be significantly beneficial in a spread of an infectious disease situation to identify potential victims from a known infected human or animal, especially when the victims are asymptomatic. Movements of objects are captured by spatio-temporal trajectories represented by a series of geospatial locations and corresponding timestamps. A large amount of spatiotemporal trajectory data is being gathered by various location acquiring sensor devices by tracking movement behaviours of people, animals, vehicles and natural events. Trajectory data mining techniques have been proposed to discover useful patterns to understand the behaviours of spatio-temporal trajectories. One unexplored pattern is to identify contacts of targeted trajectory in spatio-temporal trajectories, which is defined as contact mining. The aim of this study is to investigate contact mining from spatio-temporal trajectories. The approach will be initiated by preprocessing spatio-temporal data and then by investigating a robust contact mining framework to efficiently and effectively mine contacts of a trajectory of interest from a given set of trajectories. Experimental results demonstrate the efficiency, effectiveness and scalability of our approach. In addition, parameter sensitivity analysis reveals the robustness and insensitivity of our framework.