Selecting a Feature Set to Summarize Texts in Brazilian Portuguese

Daniel Saraiva LeiteLucia Helena Machado Rino

This paper presents a novel approach to combining features for training an automatic extractive summarizer of texts written in Brazilian Portuguese. The approach aims at both diminishing the effort of classifying features that are representative for Automatic Summarization and providing more informativeness for the summarizer to decide which text spans to include in an extract. Finding a balanced set of features is explored through WEKA. We discuss several ways of modifying the feature set and show how automatic feature selection may be useful for customizing the summarizer.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: