Modeling Herds and Their Evolvements from Trajectory Data

A trajectory is the time-stamped path of a moving entity through space. Given a set of trajectories, this paper proposes new conceptual definitions for a spatio-temporal pattern named Herd and four types of herd evolvements: expand, join, shrink, and leave based on the definition of a related term flock. Herd evolvements are identified through measurements of Precision, Recall, and F-score. A graph-based representation, Herd Interaction Graph, or Herding, for herd evolvements is described and an algorithm to generate the graph is proposed and implemented in a Geographic Information System (CIS) environment. A data generator to simulate herd movements and their interactions is proposed and implemented as well. The results suggest that herds and their interactions can be effectively modeled through the proposed measurements and the herd interaction graph from trajectory data.