Hierarchical chunking-phrase based translation

We present a statistical machine translation system which uses hierarchical chunking phrases (HCPB). The system can be seen as combination with fundamental ideas from both syntax-based translation and phrase-based translation, because the model not only complies with formal synchronous context-free grammar but also learned partial parsing knowledge CRF (conditional random fields) method. The decoder for HCPB MT system is based on chart-based CKY algorithm , and integrates N-gram language model effectively. Currently, this paper is focused on the research of Chinese-English spoken language translation. According to our final experiments, our method achieves a higher accuracy in the final results with BLEU and NIST score in 1WSLT2005.