Tagsets and Datasets: Some Experiments Based on Portuguese Language

Freitas, Cláudia; Trugo, Luiza F.; Chalub, Fabricio; Paulino-Passos, Guilherme; Rademaker, Alexandre

doi:10.1007/978-3-319-99722-3_46

Cited by 3 publications

(4 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Several scientific articles report the use of POS tagging in Brazilian Portuguese (pt-BR) journalistic texts (Fernandes et al, 2014;Freitas et al, 2018;Dos Santos and Zadrozny, 2014;de Sousa and Lopes, 2019). Dos Santos and Zadrozny (2014) performed POS tagging using a neural network extended from Collobert's work (Collobert, 2011), associating character-level representation with the usual word representations.…”

Section: Introductionmentioning

confidence: 99%

“…It also achieved good results on unseen data. Freitas et al (2018) created two corpora: The Mac-Morpho-UD (original Mac-Morpho corpus converted to the Universal Dependencies tagset) and the Mac-Morpho-UD+ PCP, which is the same as the previous with the PCP tag (past participle). Their work consists of two experiments: (1) verify the impact of tagsets on system performance and (2) verify the impact of size and quality in training using the Mac-Morpho-UD and the UD-Portuguese-Bosque.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Defining a state-of-the-art POS-tagging environment for Brazilian Portuguese clinical texts

Oliveira

Gumiel

et al. 2020

Res. Biomed. Eng.

View full text Add to dashboard Cite

Purpose Natural language processing techniques are essential for unlocking patients' data from electronic health records. An important NLP task is the ability to recognize morphosyntactic information from the texts, a process called part-of-speech (POS) tagging. Currently, neural network architectures are the state-of-the-art method, although there is a lack of studies exploiting this approach within Brazilian Portuguese clinical texts. The objective of this study is to define a state-of-the-art POS-tagging environment for Brazilian Portuguese clinical texts. Methods We reviewed multiple neural network-based POS-tagging algorithms, and the Flair tool was selected due to its exceptional performance in the journalistic domain, as there is any specific algorithm to Portuguese clinical texts. We executed a normalization process on available corpora from multiple domains (two journalistic, one biomedical, one clinical, and a new corpus composed of all three of these). The Flair algorithm was trained with all corpora, generating five models, which were evaluated with all domains. Results The clinical model achieved 92.39% accuracy (previous POS-tagging clinical work reached 91.5%); the biomedical model achieved 97.9% accuracy. All the models were assessed on their own test set. Conclusion We developed a new state-of-the-art modeling environment for POS tagging of Brazilian Portuguese clinical texts and achieved comparable results to other state-of-the-art studies in journalistic contexts.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Defining a state-of-the-art POS-tagging environment for Brazilian Portuguese clinical texts

Oliveira

Gumiel

et al. 2020

Res. Biomed. Eng.

View full text Add to dashboard Cite

show abstract

“…Em um contexto parecido, Freitas et al (2018) decidiram adicionar uma etiqueta, PCP, no corpus Mac-Morpho (ALUÍSIO et al, 2003) convertido para a gramática do projeto UD, com o objetivo de eliminar a mesma pouca especificação entre verbo e adjetivo nas formas participiais em certos contextos. Essa pequena modificação levou a um aumento na acurácia de um dado sistema que, sem a etiqueta PCP, conseguia acurácia de 0,9607 na atribuição de POS, e com a etiqueta alcançou 0,9624.…”

Section: Sobre Avaliação De Treebanksunclassified

“…Embora seja uma melhora sutil, dá suporte à tese de Manning de que uma seara a ser explorada para melhorar o desempenho dos anotadores está na modelagem linguística. Além disso, Freitas et al (2018) concluem também que, embora seja razoável afirmar que esquemas de anotação granulares sejam mais difíceis de serem aprendidos -como vimos, sistemas precisam generalizar a partir de exemplos categorizados em classes amplas, de modo que novas etiquetas tendem a dificultar a criação de agrupamentos robustos -, é também verdade que fenômenos pouco especificados, quando são de difícil distinção, podem se beneficiar de uma classe específica para agrupálos, como é o caso dos PCP.…”

Section: Sobre Avaliação De Treebanksunclassified