A Machine Learning Approach to Automatic Term Extraction using a Rich Feature Set

Merley da Silva Conrado, Thiago Alexandre Salgueiro Pardo, Solange Oliveira Rezende

Abstract: In this paper we propose an automatic term extraction approach that uses machine learning incorporating varied and rich features of candidate terms. In our preliminary experiments, we also tested different attribute selection methods to verify which features are more relevant for automatic term extraction. We achieved state of the art results for unigram extraction in Brazilian Portuguese.