Building a Frame-Semantic Model of the Healthcare Domain: Towards the identification of gender-based violence in public health data
Lívia Dutra, Arthur Lorenzi, Lorena Tasca Larré, Frederico Belcavello, Ely Edison da Silva Matos, Amanda Pestana, Kenneth Brown, Mariana Gonçalves, Victor Herbst, Sofia Reinach, Renato Azeredo Teixeira, Pedro de Paula, Alessandra Cristina Guedes Pellini, Cibele Sequeira, Éster Cerdeira Sabino, Fábio Leal, Mônica Conde, Regina Maura Zetone Grespan, Tiago Timponi Torrent · 2023
Public data systems gather a series of different information about Brazilian citizens. Such information is inserted in the system both via the selection of parameterized options and via open text fields. In this paper we describe the effort of modeling semantic frames for the lexicon of the healthcare domain as a means of tagging the open text fields in public health data so as to make them more easily interpretable by machine learning systems. This effort is one of the steps in a larger project aiming at using data science and machine learning techniques for the identification of territories prone to suffer from gender based-violence. The modeling effort currently covers 1,787 lexical units in the healthcare domain in Brazilian Portuguese, distributed in 29 semantic frames.