Facts2Law: using deep learning to provide a legal qualification to a set of facts

Over the course of the last year Lexum has started exploring the potential of deep learning (DL) and machine learning (ML) technologies for legal research. Although these projects are still under the umbrella of Lexum's research and development team (Lexum Lab, https://lexum.com/en/ailab/), concrete applications have recently started to become available. This demo focuses on one of these applications: Facts2Law.

The project benefits from a combination of factors. First, the millions of legal documents available in the CanLII database in parsable format along with structured metadata constitute a significant dataset to train AI algorithms. Second, Lexum has direct access to the knowledge and experience of one of the leading teams in AI and deep learning worldwide at the Montreal Institute for Learning Algorithms (MILA) of the University of Montreal. Third, the availability of computer engineers with cutting-edge expertise in the specifics of legal documents facilitates the transition from theory to practical applications.

Regarding concrete outcomes, Lexum's Facts2Law can predict the most relevant sources of law for any given piece of text (incorporating legal citations or not).