Agreement between Word Alignments and Phrase Sample Clauses

Agreement between Word Alignments and Phrase. Segmentations Our framework is very general and can be extended to include the agreement between word alignment and other latent structures such as phrase segmentations. The words in a Chinese sentence often con- stitute phrases that are translated as units in English and vice versa. Inspired by the alignment consistency constraint widely used in translation rule extraction (Xxxxx et al., 2003), we make the following assumption to impose a structural agreement constraint between word alignment and phrase segmentation: source words in one source phrase should be aligned to target words belong- ing to the same target phrase and vice versa. → For example, consider the C E alignment in Figure 2. We segment Chinese and English sen- tences into phrases, which are sequences of con- secutive words. Since “2002” and “APEC” belong to the same English phrase, their counterparts on joint alignment and segmentation model from unlabeled data in an unsupervised way. ≡ Formally, given a target-language sentence f f ∈ { } ≡ J = f1, . . . , fJ , we introduce a latent variable b bJ = b1, . . . , bJ to denote a phrase segmen- tation. Each label bj B, I, E, S , where B denotes the beginning word of a phrase, I denotes the internal word, E denotes the ending word, and S denotes the one-word phrase. Figure 2 shows the label sequences for the sentence pair. We use a first-order HMM to model phrase seg- mentation of a target sentence: Σ