Enhancing Risk Assessment in Legal Documents through Advanced Machine Learning

Pusarla, Sai Teja (2024) Enhancing Risk Assessment in Legal Documents through Advanced Machine Learning. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

In this thesis, we will conduct research to examine the use of BERT, Legal-BERT and Isolation Forest models to strengthen the risk assessment and anomaly detection of these legal documents. It shows that Legal-BERT, a BERT model pretrained on legal texts can outperform a general BERT model well beyond chance level with a test accuracy of 76.47%, and excellent specificity in precision, recall and F1 score measures. In addition, the Isolation Forest Algorithm, an unsupervised learning model, was able to find 975 anomalies among 19,501 clauses, and we can say that it is able to find deviants from normal threads of law. Most notably, real practice was the proving ground for these models, as they were tested in real-time on new legal texts and their applicability was confirmed. These findings support the opportunity advanced machine learning models provide for automation, increased accuracy, and scale in the analysis of legal documents and constitute a meaningful step toward automated risk assessments applied to real-world legal scenarios within the field of legal informatics.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Basilio, Jorge UNSPECIFIED
Uncontrolled Keywords:	BERT; Legal BERT; Clause; Risky; Not Risky; Isolation Forest; Anomaly
Subjects:	H Social Sciences > HD Industries. Land use. Labor > HD61 Risk Management K Law > K Law (General) Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	04 Sep 2025 10:46
Last Modified:	04 Sep 2025 10:46
URI:	https://norma.ncirl.ie/id/eprint/8778

Actions (login required)

View Item