Document type classification for Brazil’s supreme court using a Convolutional Neural Network

Silva, Nilton; Braz, Fabricio Ataides; Campos, Teófilo de; Guedes, André Luiz Pires; Mendes, Danilo; Bezerra, Davi; Gusmão, D. B.; Chaves, Felipe Borges S.; Ziegler, Gabriel; Horinouchi, Lucas Hiroshi; Ferreira, Marcelo J. P.; Inazawa, Pedro; Coelho, Victor; Fernandes, Ricardo Augusto Souza; Peixoto, Fabiano Hartmann; Filho, Mamede Maia; Sukiennik, Bernardo Pablo; Rosa, Lahis; Silva, Roberta; Junquilho, Tainá Aguiar; Carvalho, Gustavo

doi:10.5769/c2018001

Cited by 37 publications

(18 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Preocupados com o alto gasto de recursos para análise manual desses documentos, os autores utilizam de métodos computacionais, abordando estratégias baseadas em bag-of-words e Tf-Idf associadas a classificadores como SVM, além de métodos baseados em redes neurais. Em [2], a mesma base de dados é utilizada, entretanto com foco na aplicação específica de uma CNN para esse problema.…”

Section: Trabalhos Correlatosunclassified

See 1 more Smart Citation

Processamento de linguagem natural aplicado à classificação de decretos administrativos brasileiros

Ribeiro¹,

Araújo²,

Oliveira³

et al. 2020

Anais Estendidos Do XXVI Simpósio Brasileiro De Sistemas Multimídia E Web (WebMedia 2020)

View full text Add to dashboard Cite

Decretos administrativos são emitidos pelo Poder Executivo com o objetivo de controlar despesas, agências, ministérios e planos do governo. Apesar de serem textos de importância estratégica, no Brasil não existem classificações agregadas que explicitam as áreas de influência de cada um desses documentos. Sendo assim, neste trabalho apresentamos duas metodologias de classificação e sua automatização através de algoritmos clássicos e do estado da arte em aprendizado de máquina para modelagem textual. Nossos resultados explicitam desempenho de cerca de 80% de F1-score para alguns desses modelos, além de apresentarem uma visão geral desse tipo de dado e suas possibilidades e culminarem com a ampliação da transparência das ações e decisões do Poder Executivo Federal.

show abstract

Section: Trabalhos Correlatosunclassified

“…A rede convolucional utilizada parte do trabalho de [2], que demonstrou sucesso na classificação de documentos jurídicos em língua portuguesa com essa arquitetura.…”

Section: Rede Convolucionalunclassified

Processamento de linguagem natural aplicado à classificação de decretos administrativos brasileiros

Ribeiro¹,

Araújo²,

Oliveira³

et al. 2020

Anais Estendidos Do XXVI Simpósio Brasileiro De Sistemas Multimídia E Web (WebMedia 2020)

View full text Add to dashboard Cite

show abstract

“…The authors, Wei et al [4] reported the comparison of deep learning methods and SVM algorithm for four data sets, among which CNN is stated as a powerful tool which can produce better results for large volume of data sets. Silva et al [5] classified the Brazilian court system documents from the PDF datasets and conveyed all challenges during the classification process. Here, the authors used simple CNN method and obtained an accuracy of 90% on this original datasets.…”

Section: A Legal Search Datamentioning

confidence: 99%

“…Convolution neural network [4] Real datasets four CNN out performed SVM Word embedding with Neural network. [5] Brazilian court dataset 90%…”

Section: Legal Perspective Interfacementioning

confidence: 99%

An Artificial Intelligence based Analysis in Legal domain

2019

IJITEE

View full text Add to dashboard Cite

Around the world, legitimate information and common laws are available in raw form, but hard to understand and not in organized form. All legitimate information is nowadays computerized since the legal information gets generated on a regular basis in a huge volume due to increase of maritime (law) courts. The automation tool to analyse this legal data can serve effectively for lawyers and law students, which can address a lawyer’s role and can even become powerful to release such a role in future. The machine learning and deep learning algorithmsbased analysis systems apply these methods mainly for document classification. Legal document translation, text classification, summarization, data forecasting and data obtainment are part of the goals got from research charity. In this study, we review about the different methods of deep learning used in legal tasks such as Legal data search, Legal document analytics, and Legal perspective interface. To solve aggregate tasks, one can use the deep learning methods like, Recurrent Network Networks (RNN), Gated Recurrent unit network (GRU), Long Short Term Memory networks (LSTM), Convolutional neural network (CNN). Through this review, we instituted that deep learning models are giving advanced performance

show abstract

“…Furthermore, the average processing time reaches more than seven years in some cases. Such long waiting times negatively impact Brazil's legal certainty and brings about greater budgetary needs-Brazil spent R$ 90.7 billions in 2017 to maintain the judiciary, corresponding to about 28 billion 3 dollars [2].…”

Section: Introductionmentioning

confidence: 99%

Topic Modelling Brazilian Supreme Court Lawsuits

Araujo

Campos

2020

Frontiers in Artificial Intelligence and Applications

View full text Add to dashboard Cite

The present work proposes the use of Latent Dirichlet Allocation to model Extraordinary Appeals received by Brazil’s Supreme Court. The data consist of a corpus of 45,532 lawsuits manually annotated by the Court’s experts with theme labels, a multi-class and multi-label classification task. We initially train models with 10 and 30 topics and analyze their semantics by examining each topic’s most relevant words and their most representative texts, aiming to evaluate model interpretability and quality. We also train models with 30, 100, 300 and 1,000 topics, and quantitatively evaluate their potential using the topics to generate feature vectors for each appeal. These vectors are then used to train a lawsuit theme classifier. We compare traditional bag-of-words approaches (word counts and tf-idf values) with the topic-based text representation to assess topic relevancy. Our topics semantic analysis demonstrate that our models with 10 and 30 topics were capable of capturing some of the legal matters discussed by the Court. In addition, our experiments show that the model with 300 topics was the best text vectoriser and that the interpretable, low dimensional representations it generates achieve good classification results.

show abstract

Document type classification for Brazil’s supreme court using a Convolutional Neural Network

Cited by 37 publications

References 0 publications

Processamento de linguagem natural aplicado à classificação de decretos administrativos brasileiros

Processamento de linguagem natural aplicado à classificação de decretos administrativos brasileiros

An Artificial Intelligence based Analysis in Legal domain

Topic Modelling Brazilian Supreme Court Lawsuits

Contact Info

Product

Resources

About