A Non-prescriptive Environment to Scaffold High Quality and Privacy-aware Production of Open Data with AI

Data quality is strictly related to fitness for use. Therefore, data providers should improve the intrinsic quality of published data to prevent the diffusion of data sets practically impossible to use. Among all data providers, it is critical that governments and public agencies better assess and improve the quality of the produced data sets as early as possible - ideally during the production phase. Besides the quality aspect, data providers should also bear in mind that if they want to expose personal information, data must be compliant with EU General Data Protection Regulation (GDPR). Thus, this paper introduces a methodology to scaffold the agile production of Open Data (OD) that takes into account both quality and privacy concerns. To guarantee efficiency, our methodology is based on a pattern matching approach enhanced by artificial intelligence (AI) - in particular by using decision trees. The presented approach has been tested on real data sets already published as OD. In the evaluation, the advantages offered by the support of AI are evident. The described methodology is integrated into a social platform which is non-prescriptive as it supports (and does not force) users in producing better quality/privacy OD.