Court decisions in the field of tension between transparency and privacy

The research project included four work packages. Work Package 1 elaborated legal analyses based on the theoretical framework of traditional legal methods and on comparative law methods. Work Package 2 explored the technical possibilities of anonymisation and de-anonymisation based on computer science methods, namely Natural Language Processing (NLP) methods. Work Package 3 integrated the opinion of different stakeholder groups, using conventional methods of social and administrative sciences. And Work Package 4 led to the synthesis and implementation of the research results.

The justice system faces a challenge posed by the contradiction between the public interest in transparency (specifically the right of access to court decisions) and the human right to privacy (in particular the right to be forgotten). This antagonism raises fundamental questions regarding access to judgments. By linking publicly available information on the internet, data mining approaches such as machine learning can make it possible to identify the parties and thus attribute certain personal information (crimes committed, divorce, inheritance claims, political activism, etc.) to individuals, even from anonymised court decisions.

The aim of this research project was to find appropriate ways for courts to grant access to judicial documents while adequately safeguarding the right to privacy. The results should serve as guidelines for politics and the judiciary on the anonymisation of judgments.

The project contributed in two key areas within Natural Language Processing (NLP), focusing on legal texts: resource curation and detailed model analysis. For court practice, the project provides in-depth knowledge and guidelines regarding the anonymisation of court judgments. With the results achieved, the research in the field of anonymisation of court decisions is currently exhaustively covered.

Drei Hauptbotschaften

1. The courts in Switzerland have at least at the federal level far-reaching rules on the publication and anonymisation of judgements.
2. The risk of de-anonymisation is rather low at the moment.
3. The rules of anonymisation and anonymisation practice must be reviewed periodically due to differentiated findings on anoymisation requirements and ongoing technical developments.

Open Justice versus Privacy