Dataset Information

Using deep learning to predict outcomes of legal appeals better than human experts: A study with data from Brazilian federal courts.

ABSTRACT: Legal scholars have been trying to predict the outcomes of trials for a long time. In recent years, researchers have been harnessing advancements in machine learning to predict the behavior of natural and social processes. At the same time, the Brazilian judiciary faces a challenging number of new cases every year, which generates the need to improve the throughput of the justice system. Based on those premises, we trained three deep learning architectures, ULMFiT, BERT, and Big Bird, on 612,961 Federal Small Claims Courts appeals within the Brazilian 5th Regional Federal Court to predict their outcomes. We compare the predictive performance of the models to the predictions of 22 highly skilled experts. All models outperform human experts, with the best one achieving a Matthews Correlation Coefficient of 0.3688 compared to 0.1253 from the human experts. Our results demonstrate that natural language processing and machine learning techniques provide a promising approach for predicting legal outcomes. We also release the Brazilian Courts Appeal Dataset for the 5th Regional Federal Court (BrCAD-5), containing data from 765,602 appeals to promote further developments in this area.

SUBMITTER: Jacob de Menezes-Neto E

PROVIDER: S-EPMC9333285 | biostudies-literature | 2022

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Using deep learning to predict outcomes of legal appeals better than human experts: A study with data from Brazilian federal courts.

Jacob de Menezes-Neto Elias E Clementino Marco Bruno Miranda MBM

PloS one 20220728 7

Legal scholars have been trying to predict the outcomes of trials for a long time. In recent years, researchers have been harnessing advancements in machine learning to predict the behavior of natural and social processes. At the same time, the Brazilian judiciary faces a challenging number of new cases every year, which generates the need to improve the throughput of the justice system. Based on those premises, we trained three deep learning architectures, ULMFiT, BERT, and Big Bird, on 612,961 ...[more]

PMID: 35901178

Similar Datasets

Project description:Expert scientific knowledge, including medical knowledge, is relevant for the legal profession and can strongly influence rulings and sentencing in criminal law, civil law, and insurance law. The way in which this medical evidence is understood and evaluated thus has an impact both on individuals and on society as a whole. It enters legal procedures in various forms, for example, as expert witness statements and/or as a legal expert's own acquired medical knowledge. On the other hand, a legal expert may be confronted with expert medical opinions that differ in quality or content and thus have to decide which ones to follow. The aim of this qualitative study was to investigate legal experts' perceptions, experiences, and narratives regarding medical knowledge, particularly the skills and general knowledge used in their branch of the legal profession. A total of 51 semi-structured interviews with judges and prosecutors from different courts of law and from the public prosecutor's office in six different German-speaking (Zurich, Luzern, Aagrau, Obwalden, Nidwalden, Zug) and German/French-speaking (Bern) cantons of Switzerland were conducted, coded, and analyzed using Nvivo. We used a comparison thematic approach identifying common and new themes related to the research aims. Our findings suggest that Swiss judges and prosecutors believe that possessing and developing the skills and terminology required for processing medical information is important but complex, and time-consuming for their work. Additionally, several legal experts reported that their understanding of medical evidence was limited or even non-existent. Moreover, the acquisition of skills related to the assessment of medical reports and forensic evidence appeared to be unstructured. Participants reported having no formal instruction in how to evaluate or deal with medical knowledge. The sources they used to answer questions arising appeared to be in part problematic and non-standardized (internet, newspapers, etc.). Medical literature from peer-reviewed journals was used only rarely. The findings from this study suggest that law departments might wish to evaluate whether their graduates are sufficiently equipped with scientific literacy skills and appropriate skills to evaluate medical information for their later careers. At the same time, medical knowledge pertinent to forensics published in local legal journals may be more effective in reaching the legal expert audience than in medical journals.

Dataset Information

Using deep learning to predict outcomes of legal appeals better than human experts: A study with data from Brazilian federal courts.

Publications

Using deep learning to predict outcomes of legal appeals better than human experts: A study with data from Brazilian federal courts.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets