An Approach for Spatial Pattern Matching in Simulation Trajectories

For models that include spatial aspects, the description and recognition of spatio-temporal patterns is an important building block for the analysis of simulation trajectories. We propose an approach that makes use of user-definable qualitative spatial relations between moving entities to represent simulation trajectories as directed labeled graph. In this graph, spatio-temporal patterns can be found through a graph pattern matching algorithm. We implemented the approach using the graph database Neo4j and successfully tested it on movement data from the Robocup Soccer Simulation as well as spatial cell biological simulations.