Sampling-based multilingual alignment
Résumé
We present a sub-sentential alignment method that extracts high quality multi-word alignments from sentence-aligned multilingual parallel corpora. Unlike other methods, it exploits low frequency terms, which makes it highly scalable. As it relies on alingual concepts, it can process any number of languages at once. Experiments have shown that it is competitive with state-of-the-art methods.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...