Modelling Data Environments within Prov to Assist Decision Making for Anonymisation

The Anonymisation Decision-making Framework (ADF) operationalises the risk management of data exchange between organisations, referred to as ”data environments”. The overarching goal of the ADF is to provide decision makers with semi-automatic information to make good sharing decisions. Using a use case that showcases how ADF is employed, we identify how a provenance formalism could be utilised to provide a representation of information required by the ADF. From this, we identify a currently unmet requirement which is the modelling of data environments. We show how data environments can be implemented within the provenance standard W3C PROV in four different ways, and analyse the costs and benefits of each approach. We then assess the ability of PROV to model the ADF’s data situation evaluation tool. We conclude that a very large overlap exists between the ADF and PROV of the required concepts and archictecture, and the gap can be managed through modelling choices and extensions permitted within PROV. Thus, in this work, we show that PROV is a suitable machine-interpretable format for information required to make sharing and anonymisation decisions.