Recognizing Vessel Movements from Historical Data

How can movements of vessels at sea be recognized and be related to features of vessels? In this study we focus on data from AIS (Automatic Identification System) and we use historical AIS data as the basis for recognition and we extend this with prior knowledge in the form of an ontology. We present solutions for three problems. First we show that a form of piecewise linear segmentation allows strong compression of AIS data without much loss of important information. Second, we show that similarities between movements can well be measured using a form of edit-distance. These similarities are then used to predict properties of vessels and to discover clusters of vessel movement. Finally we show how knowledge that is represented in an ontology can be used in knowledge-intensive similarity which produces better results than using only movement data.