Predicting destinations and arrival times of vessels could improve the efficiency of the naval transportation industry as well as the traffic safety. While the Automatic Identification System (AIS) messages contain a dedicated destination field, this one is more than often wrongly filled. However, as we show in this paper, it still contains relevant information should it be properly processed. We propose an approach to identify, quantify and fix the errors in the AIS destination field. A "cleaning-matching" algorithm matches the firstly cleaned raw AIS message with a standard list of port names, and returns the most likely destination(s) it is referring to. The uncertainty and imprecision of the algorithms’ output is represented by belief functions. The accuracy of the algorithm is evaluated using a sample of one month of AIS data labeled with the ground truth destination. We show that the proposed algorithm increases the accuracy of the predicted destination based on the AIS destination field by 20%.
[1]
A. Harati-Mokhtari,et al.
Automatic Identification System (AIS): Data Reliability and Human Error Implications
,
2007,
Journal of Navigation.
[2]
Clément Iphar,et al.
Formalisation of a data analysis environment based on anomaly detection for risk assessment : Application to Maritime Domain Awareness
,
2017
.
[3]
Glenn Shafer,et al.
A Mathematical Theory of Evidence
,
2020,
A Mathematical Theory of Evidence.
[4]
Niklas Lavesson,et al.
Open data for anomaly detection in maritime surveillance
,
2013,
Expert Syst. Appl..
[5]
Arthur P. Dempster,et al.
A Generalization of Bayesian Inference
,
1968,
Classic Works of the Dempster-Shafer Theory of Belief Functions.