Word Alignment Based on Multi-Grain Model

Phrase Alignment Based on Combination of Multiple Strategies

“…①All non-leaf syntactic nodes whose heights are t in CT, are collected into the set A. ②SP and δ of nodes in set A are computed respectively according to formula (4) and formula (5). ③If there is node whose δ is larger than 0, then IsExist is set to 1.…”

“…CT t1 [1,3] CT t2 [2,5] CT t3 [4,7] CT t4 [6,9] e 1 e 3 e 4 e 6 e 7 e 8 e 9 e 2 e 5 The correspondence between Chinese parsing tree and English parsing tree is shown in Figure 2. This kind of phrase translation pairs is very useful in machine translation, bilingual lexicography, and translation ordering model.…”

“…An English-Chinese word alignment model based on bilingual lexicon and language knowledge is given, and it is built on the theory of formal optimal partition of the bilingual sentence pairs [4]. An approach of word alignment based on multi-grain model is proposed, where a bilingual sentence pair is split into blocks in different grains, and word alignments within each corresponding blocks are extracted, which will restrict the searching space of word alignment in the relatively accurate range and reduce the mapping errors [5]. A bootstrapping frame is designed, where the bilingual corpus is aligned based on translation dictionary, and translation dictionary is expanded based on word alignment results.…”