A graph-based data mining approach to preventing financial fraud: a case study

In this paper, we present a graph-based approach to a data mining problem of exploring and revealing domain groups of users prone to committing financial fraud. Data mining in financial applications refers to extracting and organizing knowledge from large amount of legal and financial data according to certain criteria. In order to solve this problem, information about the companies should be well-defined and arranged according to a data mining process model. Here, we introduced a graph-based model to formalize large amounts of data as well as a methodology of graph centers of interest to solve classification and prediction data mining tasks that are vital to handle fraud detection. A graph-based model consists of a set of real objects, such as shareholders, vendors, and directors, with some object attributes and relations between the objects. IBM i2 software is used to visualize data and graph model representation.