Gloss to Text Machine Translation

• Grant number: 25/03814-6
• Support Opportunities: Scholarships in Brazil - Doctorate
• Start date: May 01, 2025
• End date: April 30, 2029
• Field of knowledge: Physical Sciences and Mathematics - Computer Science
• Principal Investigator: José Mario de Martino
• Grantee: José Carlos Ferreira
• Host Institution: Faculdade de Engenharia Elétrica e de Computação (FEEC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Associated research grant: 24/00914-7 - Center for Science for Development - Assistive Technology and Accessibility in Libras, AP.CCD

Abstract

The main focus of the research to be developed by the fellow is developing a computational model based on machine learning for the automatic translation of the glossed representation of Libras into text in Portuguese. Models based on machine learning techniques for automatic translation between written languages will be studied, evaluated, and adapted for translation between the glossed representation and Portuguese. In addition to this primary focus, the PhD student will also contribute to the construction of the database and evaluation of translation approaches. The researchers will also participate in the project's Workshops. The doctoral student involved in the Glosa-to-Texto automatic translation will carry out the following activities: (i) complete the credits in courses required by the graduate program; (ii) in-depth literature review in the areas of automatic translation and machine learning; (iii) collection and preparation of data for training the models; (iv) definition, implementation, adaptation and refinement of models based on machine learning; (v) conducting experiments and evaluating the models; (vi) preparation of technical reports and scientific articles; (vii) preparation and defense of a qualification document; and (viii) preparation of the doctoral thesis.