A Study On The Translation Of Naxi Chinese Machine Based On The Integration Of Semantic Roles

The translation of Chinese language and Naxi language have a value of Naxi language research between academic research and practical applications. With the rapid development of machine translation technology, syntax-based statistical machine translation machine translation gradually become mainstream. Naxi language has its own language features, and there is a big difference between the Naxi language and Chinese grammar. Naxi language sentence the semantic role to some extent, reflect the semantic relationships between words in a sentence between these semantic relationships bilingual translators have certain constraints role, therefore, on the basis syntactic dependency tree to string model based on the integration semantic role features, explore the Naxi-Chinese machine translation method, the main focus of Naxi semantic role labeling and integration of semantic roles Naxi Chinese machine translation method. Specific research results are as follows:(1) The Naxi language semantic role labeling study based on dependency parsing.To identify the Naxi sentence semantic roles is proposed, it was based on semantic role Naxi language dependency parsing tagging methods. Analysis Naxi semantic roles, semantic role definition Naxi classification system on the basis of the Naxi language dependency parsing based on semantic role annotation artificial training set, to train a maximum entropy classification model to achieve Naxi semantic role identification. Experimental results show that identify the Naxi semantic roles have a good effect.(2) The method for the integration of semantic features of Naxi-Chinese of machine translation methods.In order to effectively utilize the Naxi characteristics of the method statement and semantic role information, it presents a fusion of semantic role features Naxi-Chinese machine translation method, it was based on dependency tree to string translation model, build a semantic role features model to obtaining semantic roles reordering probability, using the commonly used features and semantic role feature model integrated into the parameters of the log-linear model, the model is trained by minimizing the error rate. In the decoding process according to semantic role features model to adjust the relative order of the target string. Experimental results show that the integration of semantic role features Naxi Chinese machine translation methods improve the accuracy of the translation effectively.(3) The integration of semantic features of Naxi-Chinese machine translation system implementation.On the basis of the above work, and with the open-source tools, and tools have been implemented in this chapter, it is intended to develop the integration of semantic features of Chinese-Naxi language machine translation prototype system to achieve interoperability translation between bilingual Chinese sentences and Naxi sentences.