E09-1063 decoding stage , we deployed word lattice decoding . We evaluated our approach on
E09-1063 facilitates the utilisation of word lattice decoding . 4.3 Phrase-Based Word Lattice
D10-1041 lattice into the decoder to perform lattice decoding . The experiments are conducted
H94-1078 is useful to compare relative lattice decoding speeds from the various techniques
E09-1082 Dyer for his implementation of lattice decoding in Moses . This work was supported
D10-1041 baseline system which can support lattice decoding . We also realise a paraphrase
E09-1063 This motivates the need for word lattice decoding . The search process can be rewritten
D15-1122 model can bring more benefits than lattice decoding . Because the quality of a given
D15-1218 process ( translation ) . 2 . Lattice Decoding : ( Dyer et al. , 2008b ) suggests
D09-1115 contains all the nodes . While in lattice decoding , a translation path may skip
E14-1065 system with synthetic phrases to lattice decoding setup without synthetic phrases
D15-1163 2010 ) or by using paraphrases in lattice decoding ( Onishi et al. , 2010 ; Du et
E09-1063 first to propose the use of word lattice decoding in PB-SMT , in order to address
D15-1122 paraphrasing model outperforms lattice decoding . These results support our hypothesis
D15-1122 . Arch_LD represents that only lattice decoding is carried out using hybrid combination
D15-1122 hypothe - ses . One challenge of the lattice decoding model is that it is difficult
D15-1122 using the paraphrasing model and lattice decoding sep - arately , aiming to compare
D15-1122 that the paraphrasing model and lattice decoding can compensate for the weaknesses
D15-1122 systems . 5.1 Results LD represents Lattice Decoding ( Ma an d McKeown 2012 ) ; PARA
H01-1034 to use the same algorithm for lattice decoding , which may also provide improved
hide detail