A Meta-model for Data Quality Management Simulation

Data quality management initiatives could both help to prevent the occurrence of data defects and repair their effect. While such initiatives can reduce overall costs, they also cause costs for their development and implementation. Therefore, the overall aim is not to improve data quality by any means, but to ensure cost-efficiency. The paper proposes a meta-model for simulating data quality management, which can be used for planning of cost-efficient initiatives. MOTIVATION AND SIMULATION APPROACH Companies use data in their operating business, e.g. when they produce goods, or when they render services. The quality of the data used is always a critical factor regarding the outcome of business processes (e.g. process lead time, customer satisfaction, product quality). In order to be able to work with data of good quality, data quality (DQ) requirements need to be clearly defined. When they do so, companies need to be aware of the fact that both using poor data and creating good data brings about considerable costs [1]. Initiatives for data quality management (DQM) could help prevent or reduce occurrence of data defects. While such initiatives can reduce overall costs, they also cause costs for their development and implementation. Therefore, the overall aim is not to improve DQ by any means, but to ensure costefficiency when implementing DQM initiatives. Basically, the budget of DQM initiatives is limited by costs (i.e. monetary losses) that are caused by poor data without the expected effect of the DQM initiatives (cf. Figure 1.a). Taking a systems theory perspective, Figure 1.b illustrates these interrelatedness as a closed loop aiming at reducing costs arising from business problems and DQM initiatives. In this model, the system is constituted by business processes and by data used for doing operating business (including potential data defects and business problems), the sensor determines DQ and the costs arising from business problems, and the controller determines the scope and character of DQM initiatives. Literature covers costs of poor DQ [2] and the classification of DQ and DQM costs [1]. However, the effort to ensure context specific data quality and the balance between related costs and benefits are rarely Figure 1: Application of Closed-loop Control to Data Quality Management Controller (Cost Analysis and Budget Calculation) Sensor (Monitoring of Metrics and Business Problem Costs) System (Data, Processes, Failures, DQM Initiatives, etc.)