Adapting unsupervised semantic role labeling methods for Brazilian Portuguese

• Grant number: 11/22500-0
• Support Opportunities: Scholarships abroad - Research Internship - Master's degree
• Start date: March 01, 2012
• End date: June 30, 2012
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Principal Investigator: João Luís Garcia Rosa
• Grantee: Fernando Emilio Alva Manchego
• Supervisor: Mirella Lapata
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Institution abroad: University of Edinburgh, Scotland
• Associated to the scholarship: 10/04647-0 - Semi-supervised automatic semantic role labeling for brazilian portuguese, BP.MS

Abstract

One of the main challenges in semantic role labeling (SRL) is the development of applications for languages other than English. For Brazilian Portuguese, the necessary lexical resources are currently being developed. However, the amount of annotated data to be provided wouldn't allow for an appropriate supervised learning. Considering this data scarcity problem, semi-supervised and unsupervised methods should be explored. In this project, it is proposed the use of unsupervised learning methods which have been useful for English SRL, and their adaptation for Brazilian Portuguese. In particular, we pretend to study methods based on clustering and graph partitioning, employing the features proposed in the Master's project to which this Internship project is related. The resulting unsupervised system will be compared with the semi-supervised one developed during the Master's, in order to determine the advantages and disadvantages that each method presents for Portuguese language. (AU)