PUCRJ-PUCPR-UFMG at eHealth-KD Challenge 2021: a multilingual BERT-based system for joint entity recognition and relation extraction

• Autor(a) principal: Thiago Castro Ferreira
• Data de Publicação: 2021
• Outros Autores: João Vitor Andrioli de Souza, Lucas Ferro Antunes de Oliveira, Lucas Emanuel Silva e Oliveira, Claudia Moro Barra, Emerson Cabrera Paraíso, Yohan Bonescki Gumiel, Adriana Silvina Pagano, Giovanni Pazini Meneghel Paiva, Elisa Terumi Rubel Schneider, Lucas Pavanelli
• Tipo de documento: Artigo de conferência
• Idioma: eng
• Título da fonte: Repositório Institucional da UFMG
• Texto Completo: http://hdl.handle.net/1843/66341; https://orcid.org/0000-0003-0200-3646; https://orcid.org/0000-0002-8950-0890; https://orcid.org/0000-0003-4052-7993; https://orcid.org/0000-0002-8921-5598; https://orcid.org/0000-0003-2228-7965; https://orcid.org/0000-0002-3150-3503; https://orcid.org/0000-0002-6740-7855; https://orcid.org/0000-0003-2637-3086; https://orcid.org/0000-0003-1811-5087; https://orcid.org/0000-0002-9789-9547; https://orcid.org/0000-0001-8239-2930
• topic: eHealth; Entity recognition; Relation extraction; BERT; Deep learning; Processamento da linguagem natural (Computação); Informática na medicina

This study introduces the system submitted to the eHealthKD Challenge 2021 by the PUCRJ-PUCPR-UFMG team. We proposed a multilingual BERT-based system for joint entity recognition and relation extraction in multidomain texts. Our end-to-end multitasking model benefits from the transformer architecture, which has proved to capture better the global dependencies of the input text. Also, the use of a multilingual model contributed to our system to perform well even in the set of tests containing non-Spanish sentences. Our system ranked first in the entity recognition task and second in the Main scenario, where both tasks of entity recognition and relation extraction had to be solved. The full code of our approach and more details of the implementation are publicly available