Knowledge extraction from decision-making data for maritime navigation support

As maritime traffic density increases, providing navigation support for enhanced situational awareness and decision-making becomes critical. Extracting expert knowledge is challenging due to its subjective nature, and analyzing raw maritime data is often inefficient due to the overwhelming volume of non-critical information. This work proposes extracting decision-making process knowledge from Automatic Identification System (AIS) data to assist in navigation support. This approach balances the subjectivity of historical navigational decision information and mines critical information from raw traffic data. Specifically, decision-making point data categorized by maneuver type is collected from raw AIS data, followed by statistical analysis in terms of risk indicators and positional information. This analysis facilitates knowledge extraction, which is then applied to develop a rule-based decision-making algorithm. To validate this algorithm, a decision support system is designed in a professional navigation simulator and tested in a challenging encounter scenario by 12 participants with a nautical science background. The results indicate that the developed decision support system effectively provides early warnings for decision-making.