Swiss Court Rulings Corpus

We are in the process of building a diverse, multilingual, diachronic corpus of Swiss court decisions. These decisions are only semistructured text and are not directly fit for automatic analysis. We are now processing the text with parsers and regular expressions to extract interesting structured information such as sections, judgment outcomes, judicial persons, and citations. This information can be used to study many interesting tasks such as Legal Judgment Prediction (can we predict the outcome of a case based on the case's facts?) and Criticality Prediction (how important is a given decision?). Additionally, it enables us to study special properties of the Swiss legal system such as potential biases or irregularities, possibly leading to suggestions for improved processes. 

Because the format of the decisions differs from one court to another, we require a lot of manpower to process these texts. If you are interested in contributing (we may offer authorship for substantial contributions), please drop us an email at joel.niklaus@inf.unibe.ch. We publish at conferences in the Natural Language Processing domain such as ACL, EMNLP, COLING, etc.

The table below gives an overview of what has been done so far by whom with explanations where a given extraction task is not possible. The number of decisions (status 13/09/2021) is available in a column as an indicator of priority (the more decisions available, the higher the benefit).