Abstract
Obtaining a high-precision risk warning service, which can improve trading efficiency and legality, by reducing sampling proportion and customs clearance time dramatically is critical for cross-border trades. However, existing anti-fraudulent services are weak either in the precision or the mining capacity of discovering hidden risks. Among the reasons are incomplete data, untrustworthy resources, and old analysis models. On the basis of these observations, this article makes a combined technical solution for a risk warning service to address data resource, integrity, and mining capacity issues. The provided risk warning service is featured with a correlation analysis approach, which is advanced and efficient at addressing multisource and heterogeneous data to identify deep-seated risks with cross-border products, such as fake documents, price concealment, epidemic events, and ingredient pollution. To reveal the hidden correlation risks in cross-border clearance, a set of correlation-oriented data models and multi-attribute, multi-object, and multi-level methods are developed. The involved data sources and objects can be collected from inside businesses and public resources. Data are further structured to depict the whole portrait of a trade. The correlation analysis approach proves to be feasible and efficient in processing multisource and heterogeneous data to discover deep-seated risks with cross-border products. The risk warning service and the used correlation analysis approach have been studied and developed on the basis of a pilot project at an exit-and-entry port in Shanghai.
