Parallel phrase extraction from English-Vietnamese parallel corpora

Parallel phrases are important for some Natural Language Processing (NLP) tasks such as machine translation or cross language information retrieval. This paper proposes a novel method to extract parallel phrases from English-Vietnamese parallel corpora. In this method, we use predefined syntactic patterns and phrase translation probabilities for determining parallel phrases. The experiments are conducted on English-Vietnamese parallel corpora and have shown that our method increases 79:72% of Fscore for obtaining parallel phrases in comparison with a baseline.