Detecting shipping fraud in global supply chains using probabilistic trajectory classification

he globalization of trade puts significant pressure on effective customs compliance and supply chain intelligence by freight forwarders. Containerization and the asymmetric information provisioning of cargo negatively impact the ability to track goods in transit. Freight forwarders seek ways to improve their intelligence by applying data mining techniques to detect potential fraudulent declarations. This paper proposes a research project on the use of trajectory classification to analyze how goods are being transported between the consignor and consignee. The trajectory of cargo is expected to reflect patterns of fraud that are mainly ignored by modern fraud detection systems. Expected outcomes of the project are twofold. A framework will be built for freight forwarders to set up classifiers for the purpose of predicting fraudulent shipment trajectories. These classifiers are expected to improve the effectiveness by which customs compliance is enforced. In addition, supply chain data has the characteristics of big data and is therefore difficult to analyze. The framework is expected to contribute a new application of trajectory classification to the data mining literature and show how cargo trajectories can be efficiently classified.