2021/10/24 by Mohr Wenger, Tom Kalir, Wenger, Mohr +9
Economics, Econometrics and Finance · Social Sciences · #Artificial Intelligence in Law #Computation and Language (cs.CL) #Criminal Justice and Corrections Analysis #FOS: Computer and information sciences #Law, Economics, and Judicial Systems #Legal Education and Practice Innovations
paper · pdf · doi:10.48550/arxiv.2110.12383
openalex publication_date 2021/10/24 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
We present the task of Automated Punishment Extraction (APE) in sentencing\ndecisions from criminal court cases in Hebrew. Addressing APE will enable the\nidentification of sentencing patterns and constitute an important stepping\nstone for many follow up legal NLP applications in Hebrew, including the\nprediction of sentencing decisions. We curate a dataset of sexual assault\nsentencing decisions and a manually-annotated evaluation dataset, and implement\nrule-based and supervised models. We find that while supervised models can\nidentify the sentence containing the punishment with good accuracy, rule-based\napproaches outperform them on the full APE task. We conclude by presenting a\nfirst analysis of sentencing patterns in our dataset and analyze common models'\nerrors, indicating avenues for future work, such as distinguishing between\nprobation and actual imprisonment punishment. We will make all our resources\navailable upon request, including data, annotation, and first benchmark models.\n