RIASSUNTO
The growing number of remote sensing systems and ship reporting technologies (e.g. Automatic Identification System, Long Range Identification and Tracking, radar tracking, Earth Observation) are generating an overwhelming amount of spatio-temporal and geographically distributed data related to vessels and their movements. Research on reliable data mining techniques has proven essential to the discovery of knowledge from such increasingly available information on ship traffic at sea. Data driven knowledge discovery has very recently demonstrated its value in fields that go beyond the original maritime safety and security remits of such data. They include, but are not limited to, fisheries management, maritime spatial planning, gridding ship emissions, mapping activities at sea, risk assessment of offshore platforms, and trade indicators. The extraction of useful information from maritime Big Data is thus a key element in providing operational authorities, policy-makers and scientists with supporting tools to understand what is happening at sea and improve maritime knowledge. This work will provide a survey of the recent JRC research activities relevant to automatic anomaly detection and knowledge discovery in the maritime domain. Data mining, data analytics and predictive analysis examples are introduced using real data. In addition, this paper presents approaches to detect anomalies in reporting messages and unexpected behaviours at sea.