Addressing the Importance of Data Veracity During Data Acquisition for Risk Assessment Processes

The veracity of information (i.e., its quality of being and remaining true, accurate, and complete) is a pillar of efficient risk management. The informative capacity of the data on which the risk management process relies needs to be fully kept across the entire information pipeline in order to ensure that risk can be properly understood and managed. Unfortunately, research shows that the informative capacity of data may partially or entirely - be lost between the generation and the final use of a piece of information. This problem starts with the capture of information, where inconsistencies may already be observed between the reality of a phenomenon and the data supposedly reporting its measurement. As a consequence, this can lead to inadequate decision making when answering a risky event and, thus to a critical escalation of the situation. Such circumstances have been reported as contributing factors in several well-known large-impact accidents (e.g., Three Mile Island, 1979; BP Texas City Refinery, 2005; Deepwater Horizon, 2010) and continue to be faced in high-risk infrastructures nowadays. The multiplication of information sources made available through advances in the Internet of Things (IoT) and digital fields offers an opportunity to address this issue, as more and more data sources can be used to confirm a single fact. That way, decision-makers can better detect inconsistencies in the data used for risk analyses and apply appropriate corrective actions. However, this comes with several challenges. Firstly, conventional risk management approaches need to be rethought and restructured to enabling a dynamic updating of the risk picture as new information is made available. Secondly, they need to enable a characterization of the information quality by providing details on the level of uncertainties related to the generated risk picture. Thirdly, the data capture process needs to be properly understood in order to ensure that possible data corruption modes are correctly identified. This paper discusses the points above by focusing on the veracity of information during the capture of data for risk assessment purposes. We discuss how multiple data sources may be managed to reduce uncertainties in this phase. A case study on the presence of vegetation close to power lines illustrates the related implications.