Modelling argument structure in legal documents

Topic Description

Legal discourse is characterised as an argument between multiple agents who have different goals (the "adversarial system"), but who work within a common framework of argumentation and use of supporting evidence. Logical models have been developed of this type of reasoning and looked at formal models of legal reasoning.

This project will use Natural Language Processing techniques to attempt to identify the legal argument from transcripts of legal judgments. The main challenge will be how to identify the agents' different dialogue acts from the text, and to explain how these fit onto a model of the legal reasoning. There is also potential scope in this project to use the transcripts to improve existing dialogue models for legal reasoning.

Skills Required:

You should have a good undergraduate or masters degree in Computing or a related subject. Some understanding or experience of natural language processing would be an advantage.

Background Reading:

Evans, Roger; Piwek, Paul; Cahill, Lynne and Tipper, Neil (2008). Natural language processing in CLIME, a multilingual legal advisory system. Natural Language Engineering, 14(1) pp. 101–132. Available at:

Mochales, Raquel and Moens, Marie-Francine (2011). Argumentation mining. In Artificial Intelligence and Law 19(1), pp1-22}, Springer.

Report an error on this page