Combining Expert Knowledge and Databases for Risk Management

Correctness, transparency and effectiveness are the principalattributes of knowledge derived from databases. In current data miningresearch there is a focus on efficiency improvement of algorithms forknowledge discovery. However important limitations of data mining canonly be dissolved by the integration of knowledge of experts in thefield, encoded in some accessible way, with knowledge derived formpatterns in the database. In this paper we will in particular discussmethods for combining expert knowledge and knowledge derived fromtransaction databases.The framework proposed is applicable to widevariety of risk management problems. We will illustrate the method ina case study on fraud discovery in an insurance company