VERBO-BRASIL: a new lexical-semantic resource for Portuguese language processing

• Grant number: 11/22337-1
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Effective date (Start): June 01, 2012
• Effective date (End): May 31, 2013
• Field of knowledge: Linguistics, Literature and Arts - Linguistics - Linguistic Theory and Analysis
• Principal Investigator: Sandra Maria Aluísio
• Grantee: Magali Sanches Duran
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil

Abstract

This project aims to build a verbal lexicon with verb senses and respective semantic role sets. Such lexicon will be web browsable and is designed to be used in natural language processing tasks. This project aims, as well, to map the Portuguese verbal lexicon to the English verbal lexicon resulted from Semlink project (http://verbs.colorado.edu/semlink/) what allows to inherit Semlink's mappings between Wordnet, Verbnet, Framenet and Propbank. The corpus used as base for the proposed project is Propbank-Br, which was a result from semantic role label annotation over the Brazilian portion of Bosque corpus (Afonso et. al. 2002). The chosen editor to build this lexicon is Cornerstone (Choi et al. 2010). Building such lexicon is essential to a large scale annotation task in a corpus of Brazilian Portuguese in order to make available a robust training corpus for several natural language processing tasks, such as information extraction, question answering and machine translation.