From Dogma to Data: Exploring How Case Law Evolves

Professor Henrik Palmer will give a introductory talk on the super interesting "From Dogma to Data" project described below. The project uses natural language processing and network analysis to analyse case law.

Yannis Panagis will give a more technical talk exploring the machine learning approaches employed as part of the project. The synopsis is as follows: "At iCourts, the Centre of Excellence on International courts, at the Faculty of Law, we are carrying out some non-traditional quantitative research on International Courts. In this talk I will present the result of our efforts to use a family of unsupervised machine learning techniques, in order to try to understand the case-law content and the case-law development. The talk will present two different efforts one with Latent Dirichlet Allocation in combination with exploratory Network Analysis and one that has used a dynamic version of Non-Negative Matrix Factorisation and discuss their findings and probable shortcomings."

Suggestions for other talks/presentations related to natural language processing and mechanisation of law are welcome.

"From Dogma to Data"

The project is led by Professor Henrik Palmer and is funded by the Free Research Council. The project began December 1st 2014 and will run to November 20th 2017.

The project investigates how case law evolves, in a new empirical way. Hopefully, this will provide a better understanding of how the courts work.

When describing the current legal position of a certain area, it is common to use case law to show which laws apply to this area in particular. However, a complete analysis of the rulings from the courts are rarely or never available. This is in part, due to the massive amount of case law. It is simply impossible for one person to read, analyze and systematize all the rulings. The consequence of this is that the descriptions of legal positions often are based on subjective choices in terms of what case law is included and what is left out.

However, by starting with the newest developments in automated processing of textual contents, together with network analysis of legal texts, it has become possible to systematically analyze all rulings given by a specific court. In doing so, it is now possible to get a completely new empirical view of what types of cases a court decides, what is decisive in the cases and how the case law has evolved over time.

Based on a case study of three selected international courts, the project will create a database, which will be the base for communicating far more extensive knowledge on the way the courts work, than we are used to.

This creates opportunities for lawyers, as well as ordinary citizens, who will now be able to gather information on the courts case law much faster, than they could before. This will strengthen the rule of law in society, as well as contribute to better legal research and legal education.