Fishing in governmental data

An open data lab containing all information about citizens with regard to the government – Stefan de Konink thinks this would be a nice testing ground for detecting fraud.

For example, consider the data from North Brabant from the past five years: data from the Tax and Customs Administration, driving licences, municipal basic registration, the Employee Insurance Agency (UWV), citizen contacts with the police, vehicle registrations – in other words, all interactions between citizens and the government. Only the names have been deleted and the citizen service numbers encrypted. 30-Year-old computer scientist De Konink estimates that this would amount to about 1 terabyte of data for one province.

‘There are open datasets (e.g. Facebook, Twitter and LinkedIn), and there are closed datasets, including those of the municipality and other government bodies, but there is no testing ground in which all social data are brought together. Where this does occur, it is in the context of secret services. I think that it would be better to have this type of research conducted within an open academic environment, based on clear questions regarding what is to be investigated, with clear indicators as outcomes’.

This is not allowed under the Data Protection Act. The provision that data must be ‘bound to specific purposes’ prevents us from drawing connections between sources (e.g. driving licences and vehicle registrations). De Konink would like to receive a local exemption from this law, which would allow him to ‘go fishing’ for interesting ‘indicators’ within a closed network. Examples of such indicators would include the registration of dozens of vehicles at the same address, foundation directors receiving benefit payments, dozens of citizen service numbers linked to the same IP address and the registration of hundreds of companies in the same 20m2 office.

The data lab envisioned by De Konink would allow computer scientists to search for relationships between actual data. They could develop algorithms for filtering out vehicle thieves, benefit fraudsters and front men. Such connections do not constitute evidence – it is conceivable for someone to have 12 cars in the backyard as the result of a hobby that has got out of hand – but they could provide a starting point for investigating officers from the Fiscal Information and Investigation Service (FIOD) or the Tax and Customs Administration.

De Konink has received funding from the platform for research journalism, Transparant Nederland, to establish such a data lab. The Netherlands Press Council is also interested. ‘I’d rather monitor the government than help the government monitor me. A better Public Records Act would be helpful in this regard’.

‘I’d rather monitor the government than help the government monitor me’

They could develop algorithms for filtering out vehicle thieves, benefit fraudsters and front men.  (Photo: Sam Rentmeester)

They could develop algorithms for filtering out vehicle thieves, benefit fraudsters and front men. (Photo: Sam Rentmeester)

 

Stay informed about the research

Receive the Delft Outlook newsletter 4 times a year