Aprendizagem profunda para reconhecimento de entidades nomeadas em domínio jurídico
Autor(a) principal: | |
---|---|
Data de Publicação: | 2019 |
Tipo de documento: | Dissertação |
Idioma: | por |
Título da fonte: | Repositório Institucional da UFG |
Texto Completo: | http://repositorio.bc.ufg.br/tede/handle/tede/10276 |
Resumo: | Named Entity Recognition (NER) is a challenging Natural Language Processing task for a language as rich as Portuguese. When applied to a specific domain, the task acquires a new layer of complexity, handling a lexicon particular to the domain in question. In this work, it is studied the Legal domain, targeting specifically the Brazilian Labor Law. Architectures based on Deep Learning, with word representations based on static word embeddings and language models have shown state-of-the-art performance for the NER task. In this work it is used a model based on Deep Neural Networks, evaluating different forms of word representations. The evaluated models are applied to Portuguese language, for both Legal and general domains. To this end, language models based on the ELMo architecture were trained for both domains, as well as static word embeddings, specific for the Legal domain. In this work, it is verified the best type of pre-trained word embeddings for each domain, after performing a comparative study between the types of word embeddings applied to the NER task. For the training of the Legal domain NER models, ELMo and static word embeddings, two different corpora were produced and annotated, based on a collection of public documents from the Brazilian Labor Court. For the Portuguese general domain NER model, a new state-of-the-art result was achieved for the HAREM benchmark, with 83.22% F-Score for the selective scenario, and 78.04% for the total scenario. For the Brazilian Labor Law domain, a model with 93.81% F-Score was obtained. |
id |
UFG-2_b4a01062752a4f83f45fb13f3d67e22c |
---|---|
oai_identifier_str |
oai:repositorio.bc.ufg.br:tede/10276 |
network_acronym_str |
UFG-2 |
network_name_str |
Repositório Institucional da UFG |
repository_id_str |
|
spelling |
Silva, Nádia Félix Felipe dahttp://lattes.cnpq.br/7864834001694765Soares, Anderson da Silvahttp://lattes.cnpq.br/1096941114079527Silva, Nadia Felix Felipe daRosa, Thierson CoutoSoares, Anderson da SilvaCaseli, Helena de Medeiroshttp://lattes.cnpq.br/1573165588536766Castro, Pedro Vitor Quinta de2020-01-07T11:57:54Z2019-12-05CASTRO, P. V. Q. Aprendizagem profunda para reconhecimento de entidades nomeadas em domínio jurídico. 2019. 125 f. Dissertação (Mestrado em Ciência da Computação) - Universidade Federal de Goiás, Goiânia, 2019.http://repositorio.bc.ufg.br/tede/handle/tede/10276Named Entity Recognition (NER) is a challenging Natural Language Processing task for a language as rich as Portuguese. When applied to a specific domain, the task acquires a new layer of complexity, handling a lexicon particular to the domain in question. In this work, it is studied the Legal domain, targeting specifically the Brazilian Labor Law. Architectures based on Deep Learning, with word representations based on static word embeddings and language models have shown state-of-the-art performance for the NER task. In this work it is used a model based on Deep Neural Networks, evaluating different forms of word representations. The evaluated models are applied to Portuguese language, for both Legal and general domains. To this end, language models based on the ELMo architecture were trained for both domains, as well as static word embeddings, specific for the Legal domain. In this work, it is verified the best type of pre-trained word embeddings for each domain, after performing a comparative study between the types of word embeddings applied to the NER task. For the training of the Legal domain NER models, ELMo and static word embeddings, two different corpora were produced and annotated, based on a collection of public documents from the Brazilian Labor Court. For the Portuguese general domain NER model, a new state-of-the-art result was achieved for the HAREM benchmark, with 83.22% F-Score for the selective scenario, and 78.04% for the total scenario. For the Brazilian Labor Law domain, a model with 93.81% F-Score was obtained.Reconhecimento de Entidades Nomeadas (REN) é uma tarefa desafiadora em Processamento de Linguagem Natural, para uma língua tão rica quanto o Português. Quando aplicada em um domínio específico, a tarefa adquire uma nova camada de complexidade, por tratar de um léxico muito particular ao domínio trabalhado. O domínio estudado neste trabalho é o do Direito, voltado especificamente para a Justiça do Trabalho do Brasil. Arquiteturas baseadas em Aprendizado Profundo, com representações de palavras baseadas em vetores estáticos de palavras e modelos de linguagem, têm demonstrado um desempenho em nível de estado da arte para a tarefa de REN. Neste trabalho é utilizado um modelo baseado em Redes Neurais Profundas, avaliando diferentes formas de representação de palavras. São avaliados modelos tanto para o domínio do Direito quanto para a língua portuguesa em um contexto geral. Para tanto, foram treinados modelos de linguagem baseados na arquitetura ELMo para os dois domínios, assim como vetores estáticos de palavras específicos para o domínio do Direito. Neste trabalho também verificou-se os melhores tipos de vetores para cada domínio, a partir de uma série de análises comparativas entre os vetores aplicados na tarefa de REN. Para os treinos dos modelos de REN, ELMo e vetores estáticos do domínio jurídico foram produzidos e anotados em corpora específicos deste domínio, a partir da coleta de documentos públicos da Justiça do Trabalho do Brasil. Para o modelo de REN do domínio geral da língua portuguesa, atingiu-se um novo estado da arte no benchmark do HAREM, com 83.22% de F-Score para o cenário seletivo, e 78.04% para o cenário total. Para o domínio trabalhista brasileiro, foi obtido um modelo com 93.81% de F-Score.Submitted by Luciana Ferreira (lucgeral@gmail.com) on 2020-01-06T14:08:58Z No. of bitstreams: 2 Dissertação - Pedro Vitor Quinta de Castro - 2019.pdf: 1941412 bytes, checksum: c5467726f2cd684553e007670b8443ec (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5)Approved for entry into archive by Luciana Ferreira (lucgeral@gmail.com) on 2020-01-07T11:57:54Z (GMT) No. of bitstreams: 2 Dissertação - Pedro Vitor Quinta de Castro - 2019.pdf: 1941412 bytes, checksum: c5467726f2cd684553e007670b8443ec (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5)Made available in DSpace on 2020-01-07T11:57:54Z (GMT). No. of bitstreams: 2 Dissertação - Pedro Vitor Quinta de Castro - 2019.pdf: 1941412 bytes, checksum: c5467726f2cd684553e007670b8443ec (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Previous issue date: 2019-12-05application/pdfporUniversidade Federal de GoiásPrograma de Pós-graduação em Ciência da Computação (INF)UFGBrasilInstituto de Informática - INF (RG)http://creativecommons.org/licenses/by-nc-nd/4.0/info:eu-repo/semantics/openAccessReconhecimento de entidades nomeadasProcessamento de linguagem naturalDeep learningRedes neuraisLíngua portuguesaDireito do trabalhoNamed entity recognitionNatural language processingDeep learningNeural networksPortuguese languageLabor lawCIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAOAprendizagem profunda para reconhecimento de entidades nomeadas em domínio jurídicoDeep learning for named entity recognition in legal domaininfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/masterThesis-3303550325223384799600600600-77122667346336447683671711205811204509reponame:Repositório Institucional da UFGinstname:Universidade Federal de Goiás (UFG)instacron:UFGLICENSElicense.txtlicense.txttext/plain; charset=utf-82165http://repositorio.bc.ufg.br/tede/bitstreams/e806d6c4-db2c-4902-8b96-9ab39cba3a78/downloadbd3efa91386c1718a7f26a329fdcb468MD51CC-LICENSElicense_urllicense_urltext/plain; charset=utf-849http://repositorio.bc.ufg.br/tede/bitstreams/46d21799-f4bd-42cb-9fce-b3c76bf39e36/download4afdbb8c545fd630ea7db775da747b2fMD52license_textlicense_texttext/html; charset=utf-80http://repositorio.bc.ufg.br/tede/bitstreams/89768d0d-f90d-44f1-9e47-91a857ff8619/downloadd41d8cd98f00b204e9800998ecf8427eMD53license_rdflicense_rdfapplication/rdf+xml; charset=utf-80http://repositorio.bc.ufg.br/tede/bitstreams/656c4be1-61a5-47dc-8b9a-1abcb3c6d31c/downloadd41d8cd98f00b204e9800998ecf8427eMD54ORIGINALDissertação - Pedro Vitor Quinta de Castro - 2019.pdfDissertação - Pedro Vitor Quinta de Castro - 2019.pdfapplication/pdf1941412http://repositorio.bc.ufg.br/tede/bitstreams/e6c3772f-3f25-4ee0-9e9d-2bf5cf79126e/downloadc5467726f2cd684553e007670b8443ecMD55tede/102762020-01-07 08:57:54.537http://creativecommons.org/licenses/by-nc-nd/4.0/Acesso Abertoopen.accessoai:repositorio.bc.ufg.br:tede/10276http://repositorio.bc.ufg.br/tedeRepositório InstitucionalPUBhttp://repositorio.bc.ufg.br/oai/requesttasesdissertacoes.bc@ufg.bropendoar:2020-01-07T11:57:54Repositório Institucional da UFG - Universidade Federal de Goiás (UFG)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 |
dc.title.eng.fl_str_mv |
Aprendizagem profunda para reconhecimento de entidades nomeadas em domínio jurídico |
dc.title.alternative.eng.fl_str_mv |
Deep learning for named entity recognition in legal domain |
title |
Aprendizagem profunda para reconhecimento de entidades nomeadas em domínio jurídico |
spellingShingle |
Aprendizagem profunda para reconhecimento de entidades nomeadas em domínio jurídico Castro, Pedro Vitor Quinta de Reconhecimento de entidades nomeadas Processamento de linguagem natural Deep learning Redes neurais Língua portuguesa Direito do trabalho Named entity recognition Natural language processing Deep learning Neural networks Portuguese language Labor law CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO |
title_short |
Aprendizagem profunda para reconhecimento de entidades nomeadas em domínio jurídico |
title_full |
Aprendizagem profunda para reconhecimento de entidades nomeadas em domínio jurídico |
title_fullStr |
Aprendizagem profunda para reconhecimento de entidades nomeadas em domínio jurídico |
title_full_unstemmed |
Aprendizagem profunda para reconhecimento de entidades nomeadas em domínio jurídico |
title_sort |
Aprendizagem profunda para reconhecimento de entidades nomeadas em domínio jurídico |
author |
Castro, Pedro Vitor Quinta de |
author_facet |
Castro, Pedro Vitor Quinta de |
author_role |
author |
dc.contributor.advisor1.fl_str_mv |
Silva, Nádia Félix Felipe da |
dc.contributor.advisor1Lattes.fl_str_mv |
http://lattes.cnpq.br/7864834001694765 |
dc.contributor.advisor-co1.fl_str_mv |
Soares, Anderson da Silva |
dc.contributor.advisor-co1Lattes.fl_str_mv |
http://lattes.cnpq.br/1096941114079527 |
dc.contributor.referee1.fl_str_mv |
Silva, Nadia Felix Felipe da |
dc.contributor.referee2.fl_str_mv |
Rosa, Thierson Couto |
dc.contributor.referee3.fl_str_mv |
Soares, Anderson da Silva |
dc.contributor.referee4.fl_str_mv |
Caseli, Helena de Medeiros |
dc.contributor.authorLattes.fl_str_mv |
http://lattes.cnpq.br/1573165588536766 |
dc.contributor.author.fl_str_mv |
Castro, Pedro Vitor Quinta de |
contributor_str_mv |
Silva, Nádia Félix Felipe da Soares, Anderson da Silva Silva, Nadia Felix Felipe da Rosa, Thierson Couto Soares, Anderson da Silva Caseli, Helena de Medeiros |
dc.subject.por.fl_str_mv |
Reconhecimento de entidades nomeadas Processamento de linguagem natural Deep learning Redes neurais Língua portuguesa Direito do trabalho |
topic |
Reconhecimento de entidades nomeadas Processamento de linguagem natural Deep learning Redes neurais Língua portuguesa Direito do trabalho Named entity recognition Natural language processing Deep learning Neural networks Portuguese language Labor law CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO |
dc.subject.eng.fl_str_mv |
Named entity recognition Natural language processing Deep learning Neural networks Portuguese language Labor law |
dc.subject.cnpq.fl_str_mv |
CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO |
description |
Named Entity Recognition (NER) is a challenging Natural Language Processing task for a language as rich as Portuguese. When applied to a specific domain, the task acquires a new layer of complexity, handling a lexicon particular to the domain in question. In this work, it is studied the Legal domain, targeting specifically the Brazilian Labor Law. Architectures based on Deep Learning, with word representations based on static word embeddings and language models have shown state-of-the-art performance for the NER task. In this work it is used a model based on Deep Neural Networks, evaluating different forms of word representations. The evaluated models are applied to Portuguese language, for both Legal and general domains. To this end, language models based on the ELMo architecture were trained for both domains, as well as static word embeddings, specific for the Legal domain. In this work, it is verified the best type of pre-trained word embeddings for each domain, after performing a comparative study between the types of word embeddings applied to the NER task. For the training of the Legal domain NER models, ELMo and static word embeddings, two different corpora were produced and annotated, based on a collection of public documents from the Brazilian Labor Court. For the Portuguese general domain NER model, a new state-of-the-art result was achieved for the HAREM benchmark, with 83.22% F-Score for the selective scenario, and 78.04% for the total scenario. For the Brazilian Labor Law domain, a model with 93.81% F-Score was obtained. |
publishDate |
2019 |
dc.date.issued.fl_str_mv |
2019-12-05 |
dc.date.accessioned.fl_str_mv |
2020-01-07T11:57:54Z |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/masterThesis |
format |
masterThesis |
status_str |
publishedVersion |
dc.identifier.citation.fl_str_mv |
CASTRO, P. V. Q. Aprendizagem profunda para reconhecimento de entidades nomeadas em domínio jurídico. 2019. 125 f. Dissertação (Mestrado em Ciência da Computação) - Universidade Federal de Goiás, Goiânia, 2019. |
dc.identifier.uri.fl_str_mv |
http://repositorio.bc.ufg.br/tede/handle/tede/10276 |
identifier_str_mv |
CASTRO, P. V. Q. Aprendizagem profunda para reconhecimento de entidades nomeadas em domínio jurídico. 2019. 125 f. Dissertação (Mestrado em Ciência da Computação) - Universidade Federal de Goiás, Goiânia, 2019. |
url |
http://repositorio.bc.ufg.br/tede/handle/tede/10276 |
dc.language.iso.fl_str_mv |
por |
language |
por |
dc.relation.program.fl_str_mv |
-3303550325223384799 |
dc.relation.confidence.fl_str_mv |
600 600 600 |
dc.relation.department.fl_str_mv |
-7712266734633644768 |
dc.relation.cnpq.fl_str_mv |
3671711205811204509 |
dc.rights.driver.fl_str_mv |
http://creativecommons.org/licenses/by-nc-nd/4.0/ info:eu-repo/semantics/openAccess |
rights_invalid_str_mv |
http://creativecommons.org/licenses/by-nc-nd/4.0/ |
eu_rights_str_mv |
openAccess |
dc.format.none.fl_str_mv |
application/pdf |
dc.publisher.none.fl_str_mv |
Universidade Federal de Goiás |
dc.publisher.program.fl_str_mv |
Programa de Pós-graduação em Ciência da Computação (INF) |
dc.publisher.initials.fl_str_mv |
UFG |
dc.publisher.country.fl_str_mv |
Brasil |
dc.publisher.department.fl_str_mv |
Instituto de Informática - INF (RG) |
publisher.none.fl_str_mv |
Universidade Federal de Goiás |
dc.source.none.fl_str_mv |
reponame:Repositório Institucional da UFG instname:Universidade Federal de Goiás (UFG) instacron:UFG |
instname_str |
Universidade Federal de Goiás (UFG) |
instacron_str |
UFG |
institution |
UFG |
reponame_str |
Repositório Institucional da UFG |
collection |
Repositório Institucional da UFG |
bitstream.url.fl_str_mv |
http://repositorio.bc.ufg.br/tede/bitstreams/e806d6c4-db2c-4902-8b96-9ab39cba3a78/download http://repositorio.bc.ufg.br/tede/bitstreams/46d21799-f4bd-42cb-9fce-b3c76bf39e36/download http://repositorio.bc.ufg.br/tede/bitstreams/89768d0d-f90d-44f1-9e47-91a857ff8619/download http://repositorio.bc.ufg.br/tede/bitstreams/656c4be1-61a5-47dc-8b9a-1abcb3c6d31c/download http://repositorio.bc.ufg.br/tede/bitstreams/e6c3772f-3f25-4ee0-9e9d-2bf5cf79126e/download |
bitstream.checksum.fl_str_mv |
bd3efa91386c1718a7f26a329fdcb468 4afdbb8c545fd630ea7db775da747b2f d41d8cd98f00b204e9800998ecf8427e d41d8cd98f00b204e9800998ecf8427e c5467726f2cd684553e007670b8443ec |
bitstream.checksumAlgorithm.fl_str_mv |
MD5 MD5 MD5 MD5 MD5 |
repository.name.fl_str_mv |
Repositório Institucional da UFG - Universidade Federal de Goiás (UFG) |
repository.mail.fl_str_mv |
tasesdissertacoes.bc@ufg.br |
_version_ |
1798044378685505536 |