Uncovering illicit supply networks and their interfaces to licit counterparts through graph-theoretic algorithms
提出一种图论方法，利用有限信息构建非法供应链网络，识别其与合法供应链的接口，并通过整数线性规划和斯坦纳树变体算法找出犯罪者可能采用的不同供应链结构，以假冒鞋类案例验证有效性。
The rapid market growth of different illicit trades in recent years can be attributed to their discreet, yet effective, supply chains. This article presents a graph-theoretic approach for investigating the composition of illicit supply networks using limited information. Two key steps constitute our strategy. The first is the construction of a broad network that comprises entities suspected of participating in the illicit supply chain. Two intriguing concepts are involved here: unification of alternate Bills-of-materials and identification of entities positioned at the interface of licit and illicit supply chain; logical graph representation and graph matching techniques are applied to achieve those objectives. In the second step, we search for a set of dissimilar supply chain structures that criminals might likely adopt. We provide an integer linear programming formulation as well as a graph-theoretic representation for this problem, the latter of which leads us to a new variant of Steiner Tree problem: Generalized Group Steiner Tree Problem. Additionally, a three-step algorithmic approach of extracting single (cheapest), multiple and dissimilar trees is proposed to solve the problem. We conclude this work with a semi-real case study on counterfeit footwear to illustrate the utility of our approach in uncovering illicit trades. We also present extensive numerical studies to demonstrate scalability of our algorithms.