This paper argues that machine learning (ML) algorithms for legal document analysis are bottlenecked by not capturing essential features like rights and duties. It proposes using the interest theory of rights and Hohfeldian taxonomy to identify these features for stratified knowledge representation in ML. A novel heuristic based on philosophy and language models is implemented to identify Hohfeldian relations ('rights-duties' vs. 'privileges-no-rights'), achieving 92.5% accuracy using Sentence-BERT on UK religious discrimination policy texts.
Publisher
Humanities and Social Sciences Communications
Published On
May 19, 2023
Authors
Ahmed Izzidien
Tags
machine learning
legal document analysis
rights and duties
Hohfeldian taxonomy
knowledge representation
Sentence-BERT
religious discrimination
Related Publications
Explore these studies to deepen your understanding of the subject.
