Advancing Legal NLP: Application of Pre-trained Language Models in the Legal Domain

Candida Maria Greco

Abstract: Pre-trained Language Models (PLMs) are advanced technologies that enable deep-learning-based solutions for natural language processing and understanding (NLP). They show exceptional results on general domain texts, but they have been used in domain-specific fields like legal and health. However, processing texts in these fields is challenging due to syntax, structure, specialized terminology, privacy, and lack of domain-specific data. The research activities during the PhD course by the author have been focused on exploring opportunities coming from PLMs in the legal domain. Specifically, the research activities include the first systematic overview of PLM-based methods for AI-driven problems and tasks in the legal sphere, and the development of PLM-based methods to analyze the texts of European constitutions that share common topics.