D11-1079 order model within dependency treelet translation ( Quirk et al. , 2005 ) . Yet
N07-1007 source treelet is then matched to a treelet translation pair , the collection of which
N07-1007 additional feature function . The treelet translation model is estimated using a parallel
D08-1077 translation . 7.2 Models The baseline treelet translation system uses all the models described
D08-1077 extension/NN ) ( 21 ) The following treelet translation pair provides a contextual translation
N06-1002 dependency tree pairs to extract treelet translation pairs , train the order model
D08-1077 subgraph of a dependency tree ; treelet translation pairs consist of source and target
D08-1077 Training The process of extracting treelet translation pairs and order templates begins
P05-1034 sentence , extract dependency treelet translation pairs , and train a tree-based
D08-1077 exactly . When matching templates to treelet translation pairs , any unaligned treelet
N06-1002 hierarchical phrase approach , treelet translation succeeds in improving the global
D08-1077 out-of-domain test data . In the treelet translation approach ( Menezes & Quirk
D08-1077 translation As a baseline , we use the treelet translation approach ( which we previously
D08-1077 rewrite rules . Also , note how treelet translation pairs capture contextual lexical
D08-1077 hombre ( x1 ) ) ( 11 ) Matching : A treelet translation pair matches an input parse iff
D08-1077 is a correspondence between the treelet translation nodes and template nodes that
D08-1077 our current extension . 4.1 The treelet translation model Sentence-specific rewrite
N07-1007 paired data is annotated using the treelet translation system : as shown in Figure 2
D08-1077 presented an extension of the treelet translation method to include order templates
D08-1077 with the alignment relation , a treelet translation pair consisting of the source
hide detail