Judicial text processing

Two projects dealing with judicial texts have been performed in collaboration with NLP Technologies:

Over the years, RALI worked closely with the LexUM when it was a group at the Centre de recherche en droit public of the Law Faculty at Université de Montréal on judicial text processing. This research was partially funded by TAPoR (Text Analysis Portal for Research) on two main research areas:

Judicial decision anonymisation

Although Web publication of judicial decisions can make jurisprudence accessible to a wider public, there are cases where the courst require that the identify of certain persons not be made public. Luc Plamondon has developed an automatic anonymisation system using the GATE linguistic development environment. The system first detects certain named entities such as person names, locations, names of corporations and then it automatically determines if these should be changed or left as is, depending to the legal restrictions prescribed by the Law.

The main issues and the state of the system are described (in French) in an article published at TALN 2004, XIe Conférence sur le Traitement Automatique des Langues Naturelles, p. 367-376, Fès, Maroc, avril 2004.

In collaboration with Société québécoise d'information juridique (SOQUIJ), a simplified version of the system, called NOME, has been developed to be used as a Microsoft Word plug-in. The user manual and the necessary files are available here.

Contact: Guy Lapalme