Understanding the movement behaviors of pedestrians, vehicles, animals and other moving objects is important in applications such as urban traffic management system, and animal monitoring systems. Knowledge of these behaviors can lead to significant improvements in such systems.
Due to innovations in location and data mining technologies, our generation is a step ahead in realizing this goal. Current location technolgies has made large movement data available while different mining algorithms have been developed to extract as much meaningful information from raw movement data.
However, most mining algorithms give little consideration or totally ignore the application context in extracting knowledge from data even when application- and user-relevance of mining results depends on the context. As a result, movement patterns produced by such algorithms may not be meaningful to the end user's for reasons of irrelevance or the difficulty in interpreting the results.
We would like to address this problem by adding semantics to movement data in the form of interactions. Since interactions can affect movement and vice versa, we would like to exploit the strong interrelationship between the two concepts in order to make movement patterns more meaningful for the users. As the type of interactions vary with the type of moving entities involved, we would like to focus on pedestrian movement behavior. In particular, we would like to investigate the interrelationship between data mining and interactions, since data mining can bring new methods in computing the interactions but also the interactions may semantically enrich trajectory datasets to be mined.