Detalhes bibliográficos
| Ano de defesa: |
2013 |
| Autor(a) principal: |
Goulart, Rodrigo Rafael Vilarreal
 |
| Orientador(a): |
Lima, Vera Lúcia Strube de
|
| Banca de defesa: |
Não Informado pela instituição |
| Tipo de documento: |
Tese
|
| Tipo de acesso: |
Acesso aberto |
| Idioma: |
por |
| Instituição de defesa: |
Pontifícia Universidade Católica do Rio Grande do Sul
|
| Programa de Pós-Graduação: |
Programa de Pós-Graduação em Ciência da Computação
|
| Departamento: |
Faculdade de Informáca
|
| País: |
BR
|
| Palavras-chave em Português: |
|
| Área do conhecimento CNPq: |
|
| Link de acesso: |
http://tede2.pucrs.br/tede2/handle/tede/5221
|
Resumo: |
This work studies Word Sense Disambiguation (WSD) in the Biomedicine domain for English language, using external knowledge sources. Among the existing proposals for the selection of a sense for an ambiguous word, there is the graph-based approach. This approach uses a metric in the evaluation of graphs containing candidates to the correct sense for the ambiguous word. In this research, a set of metrics is analyzed individually, and, based on this evaluation, we propose a hybrid model for the selection of the metrics in order to determine the most adequate metric to be employed. The model makes use of a set of features and heuristics that determine a semi-supervised solution for WSD. The results obtained with experiments show an improvement in performance and reveal new perspectives of research. The proposed model raises the hit rate to 68,48%, increasing significantly in 3,52% the rate reported in literature |
