N06-1001 which are usually produced from a phrase aligner . In practice , the task of phrase
W12-4208 phrase tables produced by the phrase aligners are used to extract semantic
N06-1001 However , we believe that a real phrase aligner may make phrase alignment quality
N06-1001 the training data using a naive phrase aligner ( NPA ) instead of resorting
W12-4208 original corpus , which we feed the phrase aligners . The resulting phrase tables
W12-4208 before it is aligned with two SMT phrase aligners . Then the aligned lemmas are
W13-2818 between the two pars - ers , the Phrase Aligner module ( PAM , Tambouratzis et
W12-0108 the translation quality . The Phrase aligner module ( PAM ) performs offline
N06-1001 entire training corpus ; also a phrase aligner is not always available . We
W12-4208 resulting predicates are aligned by phrase aligners . In both pro- cedures , the
W12-4208 lemmatized before it is aligned by a phrase aligner , and then a " deep " method
W12-4208 predicates with the help of SMT phrase aligners and then extracting semantic
W12-3901 combined use of two modules , the Phrase aligner module ( PAM ) and the Phrasing
W12-0108 to be pre-processed using the Phrase aligner module to identify word and phrase
N06-1001 possible ones . Unlike a real phrase aligner , the NPA need not wait for the
D13-1056 F1 , a gap that would make the phrase aligner ( 85.9 % ) outperform the token
W12-0108 and their annotation ; ( b ) the Phrase aligner module , which processes a bilingual
hide detail