Abstract

Modeling legal reasoning and argumentation justifying decisions in cases hasalways been central to AI & Law, yet contemporary developments in legal NLPhave increasingly focused on statistically classifying legal conclusions fromtext. While conceptually simpler, these approaches often fall short inproviding usable justifications connecting to appropriate legal concepts. Thispaper reviews both traditional symbolic works in AI & Law and recent advancesin legal NLP, and distills possibilities of integrating expert-informedknowledge to strike a balance between scalability and explanation in symbolicvs. data-driven approaches. We identify open challenges and discuss thepotential of modern NLP models and methods that integrate

 

Quick Read (beta)