AIS Event-Based Knowledge Discovery for Maritime Situational Awareness

The discovery of anomalies and, more in general, of events of interest at sea is one of the main challenges of Maritime Situational Awareness. This paper introduces an event-based methodology for knowledge discovery without querying directly a large volume of raw data. The proposed architecture analyses the maritime traffic data to detect maritime traffic patterns and events and aggregate them in an Event Map, namely a georeferenced grid. The Event Map offers visualisation capabilities and, more importantly, is used as access interface to the maritime traffic knowledge database. The proposed methodology offers real-time access to the extracted maritime knowledge, and the possibility to perform more structured queries with respect to traditional basic queries (e.g. vessel proximity within a certain distance). The proposed approach is demonstrated and assessed using real-world Automatic Identification System (AIS) data, revealing computational improvements and enriched monitoring capabilities.